The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
TESTIMONY OF VICKIE A. TILLMAN, EXECUTIVE VICE PRESIDENT, STANDARD & POOR’S CREDIT MARKET SERVICES, BEFORE THE COMMITTEE ON BANKING, HOUSING AND URBAN AFFAIRS UNITED STATES SENATE SEPTEMBER 26, 2007 Mr. Chairman, Members of the Committee, good morning. I am Vickie A. Tillman, Executive Vice President of Standard & Poor’s (“S&P”) Credit Market Services, and head of Ratings Services, our nationally recognized statistical rating organization (“NRSRO”). I appreciate the opportunity to appear before you today. I especially appreciate your invitation because I believe it is important to clarify the role of rating agencies such as S&P in the financial markets, the rigor S&P applies in fulfilling that role, and our overall record of delivering unbiased opinions on creditworthiness. To that end, I also welcome the opportunity to address some questions that have been raised about how we have served the market in the midst of unprecedented conditions in the subprime mortgage market and the credit crunch and pressure on the economy that have followed. I want to assure you at the start of my testimony that we have learned hard lessons from the recent difficulties in the subprime mortgage area. While we fully agree with Secretary Paulson’s observation last week that “the subprime mortgage market improved access to credit and homeownership for millions of Americans,” it appears that abuses may have occurred in the origination process. We support Congress’ efforts to investigate those abuses and to prevent their recurrence. For our part, we are taking steps to ensure that our ratings — and the assumptions that underlie them — are analytically sound in light of shifting circumstances. As I am sure you know, and as my testimony will set forth in some detail, S&P began downgrading some of its ratings in this area towards the end of last year and had warned of deterioration in the subprime sector long before that. Nonetheless, we are fully aware that, for all our reliance on our analysis of historically rooted data that sometimes went as far back as the Great Depression, some of that data has proved no longer to be as useful or reliable as it has historically been. Additionally, the collapse of the housing market itself has been both more severe and more precipitous than we had anticipated. As I will describe in more detail later, we have taken a number of steps in response to enhance our analytics and process and continue to look for ways in which to do still more. Our reputation and our track record are the core of our business, and when they come into question, we listen and learn. We take our work seriously, very seriously, and at no time in our history more than now, as I speak to you. In my testimony I would like to address four broad topics: • First, the nature of S&P’s ratings and their role in the capital markets; • Second, S&P’s approach to rating residential mortgage-backed securities (“RMBS”), including mortgage securities backed by subprime mortgage loans; • Third, a number of the questions that have been raised in the press and elsewhere related to ratings, including: • Questions as to whether payment of fees by issuers presents a conflict of interest that could compromise analytical independence; • Questions as to whether S&P is somehow involved in “structuring” RMBS and other structured finance transactions; • Questions about the appropriateness of our ratings because securities backed by subprime collateral sometimes receive ‘AAA’ ratings; and • Questions about whether S&P has acted too slowly in responding to the deterioration of the subprime mortgage market. 2 • Fourth, steps we have taken in light of the Credit Rating Agency Reform Act passed by this body in 2006. Ratings and Their Role In The Capital Markets I would like to begin today by discussing the nature of our credit ratings, as it appears from numerous press reports that this matter is sometimes misunderstood. At their core, S&P’s credit ratings represent our opinion of the likelihood that a particular obligor or financial obligation will timely repay owed principal and interest. Put another way, we assess the likelihood, and in some situations the consequences, of default — nothing more or less. When we issue a rating on a particular security we are expressing our view that the security shares similar credit characteristics to those securities that have, in the past, represented a particular range of credit risk. A bond that we rate as ‘BBB’ has received the lowest of our so-called “investment grade” ratings; one rated ‘BB’ has received the highest non-investment grade rating. “Investment grade” securities are those securities that certain regulated investors may legally purchase. On S&P’s ratings scale, such securities are those rated at the ‘BBB’ level or higher. Since we began rating RMBS in the late 1970’s, only 1.09% of those securities rated by us ‘BBB’ have ever defaulted. For ‘BBs’ this number is 2.11%. Thus, when we rate securities, we are not saying that they are “guaranteed” to repay but the opposite: that some of them will likely default. Even our highest rating — ‘AAA’ — is not a guarantee or promise of performance. although rarely. 3 ‘AAAs’ do default and have defaulted, Another misconception about ratings relates to their purpose and use. Ratings speak to one topic and one topic only — credit risk. As we have repeatedly made clear in public statements, including statements to the SEC, testimony before Congress, and innumerable press releases, ratings do not speak to the likely market performance of a security. Thus, ratings clearly do not address: o Whether investors should “buy”, “sell” or “hold” rated securities; o Whether any particular rated securities are suitable investments for a particular investor or group of investors; o Whether the expected return of a particular investment is adequate compensation for the risk; o Whether a rated security is in line with the investor’s risk appetite; o Whether the price of the security is appropriate or even commensurate with its credit risk; or o Whether factors other than credit risk should influence that market price, and to what extent. I want to be clear. Ratings matter; as the individual who oversees S&P’s ratings business I would be the last person to suggest to you that they do not. But in the current climate, it is especially important to bear in mind just what it is we do and that other developments also affect market perceptions and behavior. The current credit crunch is very real, but we certainly have not witnessed widespread defaults of mortgage-backed securities. This dynamic and its relationship to the nature of ratings was recently recognized by one of Europe’s top regulators, Mr. Eddy Wymeersch, Chairman of the Committee of European Securities Regulators and also Chairman of Belgium’s Banking and Financial Commission. According to Mr. Wymeersch: 4 “[t]he press and general opinion is saying it’s the fault of the credit rating agencies . . . Sorry, the ratings are just about the probability of default. Nothing more. Now we have a liquidity crisis and not a solvency crisis.” Though they may move more slowly than market prices, ratings are not designed to be static. Our view of an RMBS transaction evolves as facts and circumstances develop, often in ways that are difficult to foresee. We issue ratings and, as new information becomes available with the passage of time, we either affirm those ratings — i.e., leave them unchanged — upgrade them, downgrade them, or put them on CreditWatch, which is a warning to the market that the rating is subject to change after a pending review. To make such decisions, we perform surveillance on our ratings. I will discuss our surveillance process in greater detail a little later on, but the three important points here are: • That we have a team and process in place whose responsibility it is to monitor developments and bring about ratings changes to reflect those developments as appropriate; • Changes in RMBS ratings are not based on speculation or market sentiment; and • Such changes are often based upon events which were not predictable. To cite only a few recent examples on this last point, the level of early payment default trends in recent subprime loans is unprecedented; so is the fact that, while individuals who purchased homes have generally paid their mortgages before paying off their credit cards, that now appears no longer to be true to the extent it once was; so is the reality that, while individuals who live in homes they purchase historically repay the mortgages on these homes more regularly than those who live elsewhere, that long-standing pattern now appears of questionable validity in a striking number of cases. These are ahistorical behavioral 5 modes, ones of particular import at a time of a substantial fall in real estate prices, and ones that, together with other factors, required downgrading some RMBS ratings even though no substantial amount of pool losses have occurred. I said earlier that we have made repeated statements about the nature and role of ratings. To the extent those efforts have failed to communicate sufficiently clearly about that topic, we view this hearing, and this process overall, as an opportunity to begin to rectify that. We recognize that we bear primary responsibility for getting the message out. We are making, and will continue to make, every effort to do so. S&P’s Rating of Securities Backed By Mortgage Loans, Including Subprime Loans Our ratings of residential mortgage-backed securities, particularly RMBS backed by pools containing subprime mortgage assets, have recently received a significant amount of attention. S&P has been rating RMBS for thirty years and has developed industry-leading processes and models for evaluating the creditworthiness of these transactions. As a result, S&P has an excellent track record of assessing RMBS credit quality. For example, S&P’s cumulative U.S. RMBS default rate by original rating class (through September 15, 2007) is as follows: Initial Rating % of Default AAA AA A BBB BB B 0.04 0.24 0.33 1.09 2.11 3.34 6 Default statistics are the critical measure of ratings analytics because, as I explained earlier, at their core ratings speak to the likelihood of timely repayment, not other market factors, such as supply and demand, that may go into the pricing of securities. Moreover, these default numbers for our RMBS ratings are lower, in some cases materially lower, than the long-term default rates for similar ratings issued on corporate bonds. While evaluating the credit characteristics of the underlying mortgage pool is part of our RMBS ratings process, S&P does not rate the underlying mortgage loans made to homeowners or evaluate whether making those loans was a good idea in the first place. Originators make loans and verify information provided by borrowers. They also appraise homes and make underwriting decisions. In turn, issuers and arrangers of mortgage-backed securities bundle those loans and perform due diligence. They similarly set transaction structures, identify potential buyers for the securities, and underwrite those securities. For the system to function properly, S&P relies, as it must, on these participants to fulfill their roles and obligations to verify and validate information before they pass it on to others, including S&P. Our role in the process is reaching an opinion as to how much cash we believe the underlying loans are likely to generate towards paying off the securities eventually issued by the pool. That is the relevant issue for assessing the creditworthiness of those securities. As a practical matter, S&P’s analysis of an RMBS transaction breaks down into the following categories: The LEVELS® Model The first step in our analysis is evaluating the overall creditworthiness of a pool of mortgage loans by conducting loan level analysis using our Loan Evaluation and Estimate of Loss System (LEVELS®) Model. This model is built on, and 7 reflects, our analytical assumptions and criteria. S&P’s criteria do not dictate the terms of the mortgage loans; those terms are set by the originator in the underwriting process. S&P collects up to seventy different types of inputs, including, but not limited to: the amount of equity a borrower has in the home; the loan type; the extent of income verification; whether the borrower occupies the home; and the purpose of the loan. This analysis allows us to quantify multiple risk factors, or the layered risk, and allows us to assess the increased default probability that is associated with each factor. Based on the individual loan characteristics, the LEVELS® model calculates probabilities of default and loss realized upon default. The assumptions and analysis embedded in the LEVELS® model are under regular review and are updated as appropriate to reflect our current thinking about rating residential mortgages. As part of our commitment to transparency, S&P makes its LEVELS® model available to investors who wish to license it. The vast majority of those involved in issuing RMBS have access to LEVELS® and use it regularly. We also publicly announce any changes to our LEVELS® model in a timely manner. In other words, our basic criteria is out there every day, subject to criticism and comment. The SPIRE® Model Another important aspect of our rating process is assessing the availability of cash flow, which comes from the monthly payments generated by the mortgage loans, to timely pay principal and interest. To do this, we use our Standard & Poor’s Interest Rate Evaluator (SPIRE®) Model. The model uses the S&P mortgage default and loss assumptions (generated by the LEVELS® model) as well as interest rate assumptions. Like the LEVELS® model, our SPIRE® model reflects our analysis and assumptions and is regularly reviewed and updated as warranted. 8 Also like our LEVELS® model, our SPIRE® model is publicly available, used extensively by market participants, and subject to market comment and review every day. Review of Originator and Servicer Operational Procedures S&P also reviews the practices, polices, and procedures of the originators and servicers primarily to gain comfort with the ongoing orderly performance of the transaction. For an originator, the topics we review include, but are not limited to: loan production practices; loan underwriting; and quality control practices and findings. S&P may adjust its credit support calculation based on the underwriting employed at origination. Review of Legal Documents S&P also reviews, with the assistance of internal and external counsel, the legal documents of the securities to be issued, and, where appropriate, opinions of third-party counsel that address transfer of the assets and insolvency of the transferor, as well as security interest and other legal or structural issues. S&P reviews the underlying documentation in order to understand the payment and servicing structure of the transaction. Credit Enhancement Any description of our ratings of RMBS would be incomplete without discussing the critical concept of credit enhancement. Credit enhancement is the protection (i.e., additional assets or funds) needed to cover losses in deteriorating economic conditions, sometimes referred to as “stress”. Sufficient credit enhancement allows securities backed by a pool of subprime collateral to receive what might otherwise be considered high ratings. One form of credit enhancement, although there are several, would occur if the pool has more in collateral than it issues in securities, thereby creating a cushion in the pool. We refer to this form of credit enhancement as 9 “overcollaterization,” and it is a key component in our ratings analysis. It provides protection against defaults in the underlying securities. That is, if the pool ends up experiencing losses, it should still generate enough cash from which to pay the holders of the securities. I will discuss credit enhancement in more detail later in my testimony. The Rating Committee After reviewing the relevant information about a transaction, including information related to credit enhancement, the lead analyst then takes the transaction to a rating committee. As with all S&P ratings, structured finance ratings are assigned by committee. Committees are comprised of S&P personnel who bring to bear particular credit experience and/or structured finance expertise relevant to the rating. The qualitative judgments of committee members at all stages of the process are an integral part of the rating process as they provide for consideration of asset and transaction specific factors, as well as changes in the market and environment. Personnel responsible for fee negotiations and other business-related activities are not permitted to vote in ratings committees and vice versa. Notification and Dissemination Once a rating is determined by the rating committee, S&P notifies the issuer and disseminates the rating to the public for free by, among other ways, posting it on our Web site, www.standardandpoors.com. Along with the rating, we frequently publish a short narrative rationale authored by the lead analyst. The purpose of this rationale is to inform the public of the basis for S&P’s analysis and enhance transparency to the marketplace. Surveillance After a rating has been issued, S&P monitors or “surveils” the rating to review developments that could alter the original rating. The surveillance process seeks to 10 identify those issues that should be reviewed for either an upgrade or a downgrade because of asset pool performance that may differ from original assumptions. The surveillance function also monitors the credit quality of entities that may be supporting parties to the transaction, such as liquidity providers. Analysts review performance data periodically during the course of the transaction, and as appropriate present that analysis to a rating committee for review of whether to take a rating action. The rating committee then decides whether the rating change is warranted. For changes to public ratings, a press release is normally disseminated. S&P’s Commitment to Constant Improvement While our ratings process is the product of three decades of analytical experience and excellence, we are always looking for ways to enhance that process and our analytics. This is a hallmark S&P principle and is especially true when, as with recent subprime loans, developments indicate that historically-rooted behavioral patterns that have served as solid foundations for analysis may lack their prior value. By now there is no doubt that subprime loans made from late 2005 through at least early 2007 are behaving very differently from loans in prior periods, even when the loans share the same basic credit characteristics. For example, for years a primary indicator of a borrower’s credit has been so-called FICO credit scores. FICO scores are provided by another independent market participant and are an industry standard. In recent loans, we are seeing borrowers with high FICO scores behaving in a manner consistent with how materially lower FICO borrowers have historically behaved. Similarly, as I observed earlier, there are a number of other ahistorical anomalies that make more problematic applying a number of historically-rooted assumptions about the behavior of borrowers. At the same time, these 11 behaviorial shifts appear not to be occurring in loans generated in 2004 and most of 2005, which include many of the same type of subprime characteristics present in the more recent loans. We are still gathering data to analyze the causes for these inconsistent market dynamics. In response to these developments, and as part of our constant commitment to enhancing our analytical processes, S&P has already initiated a number of steps: • We have significantly heightened the stress levels at which we rate and surveil transactions to account for deteriorating performance as evidenced by data we have received. We have also increased the frequency of our review of rated transactions; • We are modifying (and will soon be releasing) our LEVELS® model to incorporate these new stress levels and other changes recently made to our ratings assumptions, as announced in our July 10, 2007 press release; • We recently acquired IMAKE consulting and ABSXchange. These services have long provided data, analytics and modeling software to the structured finance community and we feel they will further enhance our indepth surveillance process; • We have also undertaken a survey of originators and their practices, particularly with respect to issues of data integrity. We are in the process of compiling the results of this survey and will publish those results when finalized; and 12 • We have hired a Chief Compliance Officer to augment our internal control procedures. In addition to these steps, we continue to look at areas in which we can further enhance our analysis and processes. Some of the areas include: • Our policies and procedures to protect against conflicts of interest; • The quantity and quality of data available to us; and • Modification of our analytics to reflect changing credit behaviors. S&P’s Response To Various Questions Some have raised questions about ratings and the ratings process in recent months in light of the turmoil in the subprime mortgage market. As I have previously said, we are well aware that certain historically-rooted assumptions we made in determining which RMBS ratings to issue do not, in retrospect, appear to have remained as relevant as they previously have been. Whether that is because of factors unique to the period immediately prior to and after 2006 or whether we must change those assumptions on a long-term basis is a subject of robust and continuing examination and re-examination at S&P. At the same time, some of the questions recently put to S&P reflect a fundamental misunderstanding of what ratings are or are based on inaccurate or, in some cases, incomplete information. Let me now address those questions. The “Issuer Pays” Model Does Not Compromise the Independence and Objectivity of Our Ratings A number of commentators have asked whether payment of fees by issuers and/or their representatives presents a conflict of interest that compromises the independence and 13 objectivity of ratings. Skeptics question whether, in pursuit of fees, S&P and other major rating agencies may give higher ratings than they otherwise would. Not only is this not true at S&P, but this line of questioning ignores the significant benefits of the “issuer pays” model to the market. S&P currently makes all of its public ratings available to the market free of charge in real time. When a rating is assigned or changed, the announcement is made on our Web sites — www.sandp.com and www.ratingsdirect.com — and a press release is provided to news outlets and other media. Today there are approximately 9 million current and historical ratings available on RatingsDirect. In addition, as many as 1.3 million active ratings are available for free on www.sandp.com. The benefits to the market are obvious: any and all interested market participants can access the same information at the same time. It creates a level playing field and a common basis for analyzing risk. It also leads to higher quality ratings as our analysis is subject to market scrutiny and reaction every day from every corner of the capital markets. This type of free, public disclosure and transparency is only possible under the “issuer pays” model. Developing and maintaining models and hiring experienced and skilled analytical talent is costly. Without payment by issuers, those costs would have to be covered by subscription fees, an approach with several insurmountable problems. A subscription model would severely limit the transparency and broad (and free) dissemination of ratings, as access would necessarily be expensive and exclusive to subscribers. Not only would this result in less, not more, information in the market, but it would also take away an important check on ratings quality — the constant scrutiny of a broad market. Moreover, because 14 subscription fees would necessarily be significant (given the breadth of our ratings coverage and the depth of our analysis), many investors, including the vast majority of individual investors, simply would not be able to afford access to ratings information. The likely result would be one of two equally harmful outcomes: either (i) these investors would have no meaningful access to ratings information; or (ii) a ratings black market would develop in which S&P’s intellectual property — its ratings analysis — would be misused or resold in a manner all too consistent with the pervasive misuse of other intellectual property and with the same destructive impact. As noted, some have questioned whether the “issuer pays” model has led S&P and others to issue higher, or less rigorously analyzed, ratings so as to garner more business. First and foremost, there is no evidence — none at all — to support this contention with respect to S&P. This is not surprising since it would be clearly against S&P’s self-interest as well as its cornerstone principles. Indeed, what evidence there is on the subject shows the opposite. 1. Consider, for instance, the performance of our RMBS ratings. As reflected in the chart below, in every year from 1994 through 2006, upgrades of U.S. RMBS ratings significantly outpaced downgrades by multiple factors — about 7:1 on average. The ratio was even higher from 2001-2006. That is to say, after S&P initially provided its ratings in this area, actual performance of the rated transactions led to upgrades far more often than downgrades as time passed. 15 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 % of Ratings Upgraded 6.81 2.54 1.38 2.54 2.20 2.78 10.08 10.21 9.24 12.82 10.74 7.91 3.79 % of Ratings Downgraded 2.21 1.70 1.18 1.25 1.28 0.54 0.98 0.64 1.04 1.93 1.05 0.85 0.45 If, as some claim, S&P deliberately issued high ratings to please those who paid for them, one would expect that the initial (allegedly inflated) ratings would require downward adjustment to reflect actual performance. Similarly, one would expect default rates on our RMBS ratings to be higher — indeed, materially higher — than the statistics I cited earlier. But, over the years, the opposite has emphatically been the case. 2. Similarly, if S&P put revenue ahead of analytical rigor, we would not refuse to rate, as we have, transactions that do not meet our criteria. A recent highly publicized example occurred in Canada where significant amounts of asset-backed commercial paper became illiquid. The paper had not met S&P’s minimum criteria and so we did not rate it. These are not the actions of an agency that would rate every deal that reaches our door. 3. The primacy of our reputation has been recognized by independent sources. A report prepared by two Federal Reserve Board economists found “no evidence” that rating agencies acted in the interest of issuers due to a conflict of interest. After detailed study, the report concluded that “rating agencies appear to be relatively responsive to reputation concerns and so protect the interests of investors.” See Daniel M. Covitz & Paul Harrison, Testing Conflicts of Interest at Bond Ratings Agencies with Market Anticipation: Evidence that Reputation Incentives Dominate (Dec. 2003) at 16 http://www.federalreserve.gov/pubs/feds/2003/200368/200368pap.pdf. The real question is not whether there are potential conflicts of interest in the “issuer pays” model, but whether they can be effectively managed by S&P and other credit rating agencies. Mr. Erik Sirri, director of the SEC’s Division of Market Regulations, recently testified at a congressional hearing that the conflicts raised by this long-standing business model are indeed manageable. As Mr. Sirri testified: “Typically, [rating agencies] are paid by the underwriter or the issuer. That presents a conflict. But we believe that conflict is manageable. Credit rating agencies should have polices and procedures in place, and they should adhere to those policies and procedures when they evaluate deals.” S&P maintains rigorous policies and procedures designed to ensure the integrity of our analytical processes. For example, analysts are not compensated based upon the amount of revenue they generate. Nor are analysts involved in negotiating fees. Similarly, individuals responsible for our commercial relationships with issuers are not allowed to vote at rating committees. These policies, and others, have helped ensure our long-standing track record of excellence. As previously noted, our track record speaks for itself. Moreover, the Credit Rating Agency Reform Act of 2006, and the SEC’s implementing regulations, give greater assurance that those policies will be enforced. S&P Does Not “Structure” Transactions Similar misunderstandings have led some to question whether rating agencies “structure” transactions, thereby threatening ratings independence. These questions are particularly troubling as they give false and negative impressions about a practice that benefits the markets — the open dialogue between issuers and ratings agencies. 17 It is true that our analysts talk to issuers of RMBS transactions as part of the ratings process, as they have traditionally had discussions with corporate issuers with respect to rating their non-structured securities. This dialogue provides benefits to the marketplace. Critical to our ability to rate transactions is a robust understanding of those transactions. Reading documents and reviewing the results of modeling are important, of course, but so is communication with the people responsible for the transaction itself. Through dialogue with issuers and their representatives our analysts gain greater insight into transactions to be rated, including any modifications to those transactions that may occur as the process goes forward. This dialogue promotes transparency into our ratings process, a virtue we believe in, and one that regulators have consistently espoused. Nor does the dialogue amount to “structuring” by S&P, even in cases where the discussion is about the effect different structures may have on ratings. S&P does not tell issuers what they should or should not do. Our role is reactive. Using our models with set publicly available criteria, issuers provide us with information and we respond with our considered view of the ratings implications. In the process, and as part of our commitment to transparency, we also may discuss the reasoning behind our analysis. Those who question this practice ignore that the ratings process is not and should not be a guessing game. Without informed discussion, issuers would be proposing structure upon structure until they stumbled upon the structure that best matches with their goals. That certainly would not make the markets more transparent and efficient. Nor should anyone view as suspicious the fact that some issuers structure transactions so as to achieve a specific rating result. Indeed, a variety of potential structures could merit a 18 particular result. Our role is to come to a view as to the structures presented, but not to choose among them. Again, we do not compromise our criteria to meet a particular issuer’s goals. As noted, we make criteria publicly available. If we were not applying our criteria to particular transactions, it would be readily apparent to the market and would immediately diminish the credibility — and thus the value — of our ratings business. Credit Enhancement — How Securities Backed By Subprime Mortgages Can Receive, and Merit, Investment Grade Ratings A potentially incomplete understanding of the ratings process has also led to questions about how a pool of subprime mortgage loans can support securities with investment grade, even ‘AAA’ ratings. The answer lies in the concept of credit enhancement. As discussed earlier, credit enhancement — additional assets or funds — affords protection against losses in deteriorating conditions. When an issuer comes to us with a pool of subprime loans to be used as collateral for an RMBS transaction, S&P is well aware, of course, that all of this collateral is not likely to perform from a default perspective like ‘AAA’ securities. Nonetheless, the pool of collateral loans will yield some amount of cash, even under the most stressful of economic circumstances. A key component of our analysis is looking at the pool of collateral to determine how much credit enhancement — extra collateral, for example — would be needed to support a particular rating on the securities to be backed by that collateral. To do this, we analyze the expected performance of the collateral in stressful economic conditions. To determine the amount of credit enhancement that could support an ‘AAA’ rating, we use our most stressful economic scenario, including economic conditions from the Great Depression. The stress 19 scenarios are then adjusted for each rating category. Thus, if our analysis of a particular collateral pool’s expected performance indicates that the pool would need 30% credit enhancement to support an ‘AAA’ rating, the issuer would have to have 30% additional collateral above and beyond the value of the securities issued in order for the securities issued by the pool to have enough credit enhancement for an ‘AAA’ rating. To put it in more concrete terms, if the pool was comprised of, for example, $1.3 million in collateral, it could only issue $1 million in ‘AAA’ rated securities in this scenario. This way, if the collateral performs poorly — and thirty percent in losses is very poor performance — there will still be sufficient collateral to cover losses incurred upon loan defaults. This credit enhancement figure would, of course, be lower for ratings other than ‘AAA’, as those ratings address the likelihood of repayment in less stressful economic environments. For example, the issuer might be able to issue $1.2 million in ‘BBB’ rated securities backed by the same collateral pool. Thus, it is not the case that through securitization, poor credit assets magically become solid investments. Rather, it is because, in our example, a pool has $1.3 million in collateral to support $1 million in securities that it may receive an entirely appropriate ‘AAA’ rating on those securities. S&P Has Been Warning the Market, and Taking Action, in Response to Deterioration in the Subprime Market Since Early 2006 Others have questioned whether S&P has acted quickly enough in response to the deteriorating subprime market. Again, we believe these questions result from an incomplete understanding of the facts. S&P has spoken out — and taken action — early and often on subprime issues. 20 For some time S&P has been through our publications repeatedly and consistently informing the market of its concerns about the deteriorating credit quality of RMBS transactions. For example: • In a January 19, 2006 article entitled U.S. RMBS Market Still Robust, But Risks Are Increasing And Growth Drivers Are Softening, we said: “Standard & Poor’s expects that some of the factors that drove growth in 2005 will begin to soften in 2006 . . . . Furthermore, Standard & Poor’s believes that there are increasing risks that may contribute to deteriorating credit quality in U.S. RMBS transactions; it is probable that these risks will be triggered in 2006.” • On May 15, 2006, in an article entitled A More Stressful Test Of A Housing Market Decline On U.S. RMBS, we reported on the results of our follow-up analysis to our September 2005 housing-bubble simulation. We stated: “[t]he earlier simulation had concluded that most investment-grade RMBS would weather a housing downturn without suffering a credit-rating downgrade, while speculative-grade RMBS might not fare so well . . . . In the updated simulation . . . [S&P used] more stressful macroeconomic assumptions [which] lead to some downgrades in lower-rated investment-grade bonds.” • On July 10, 2006, in an article entitled Sector Report Card: The Heat Is On For Subprime Mortgages, we noted that downgrades of subprime RMBS ratings were outpacing upgrades due to “collateral and transaction performance.” The article also identifies “mortgage delinquencies” as a “potential hot button,” and notes that such delinquencies “may become a greater concern for lenders and servicers.” • On July 17, 2006, we noted a 38% increase in downgrades in U.S. RMBS, a significant number of which came from the subprime market. Structured Finance Global Ratings Roundup Quarterly: Second-Quarter 2006 Performance Trends. • On Oct. 16, 2006, in our Ratings Roundup: Third-Quarter 2006 Global Structured Finance Performance Trends, we reported a 15% decline in upgrades for U.S. RMBS while the number of downgrades more than tripled compared to the same period in 2005. We also noted that the quarter’s ratings actions among RMBS transactions had set a record for the most performance-related downgrades. 21 • Then on December 8, 2006, in an article entitled Credit Trends: 2007 Global Credit Strategy: Asset Class Outlook, we informed the market of our view that “[c]redit quality in the RMBS sub-prime market has been under scrutiny this year. Standard & Poor’s RMBS surveillance group sees the environment ahead as portending greater downgrade potential along with lower upgrade potential.” We also stated that “the jump in third-quarter downgrade activity for the sub-prime market raises some risk flags for this segment; with 87 third-quarter downgrades adding to the 46 downgrades of the second quarter and 34 in the first.” • On January 16, 2007, in an article entitled Ratings Roundup: Fourth-Quarter 2006 Global Structured Finance Performance Trends, we stated: “Rating activity among subprime transactions has started to shift to being predominantly negative from being predominantly positive. . . . We expect this trend in subprime rating performance to continue during 2007.” • Ten days later on January 26, 2007, in our Transition Study: U.S. RMBS Upgrades Are Down And Downgrades Are Up In 2006, we reported that for 2006 “[d]owngrades overwhelmed upgrades for subprime mortgage collateral” and that we expected “losses and, therefore, negative rating actions to continue increasing during the next few months relative to previous years.” • Our statements to the market continued throughout the first half of 2007. On March 22, 2007, in an article entitled A Comparison Of 2000 and 2006 Subprime RMBS Vintages Sheds Light On Expected Performance, we stated: “[w]hile subprime mortgages issued in 2000 have the distinction of being the worst-performing residential loans in recent memory, a good deal of speculation in the marketplace suggests that the 2006 vintage will soon take over this unenviable position.” • In an April 27, 2007 article entitled Special Report: Subprime Lending: Measuring the Impact, we stated: “The consequences of the U.S. housing market’s excesses, a topic of speculation for the past couple of years, finally have begun to surface. . . . Recent-vintage loans continue to pay the price for loosened underwriting standards and risk-layering in a declining home price appreciation market, as shown by early payment defaults and rising delinquencies.” • Then on June 26, 2007, in an article entitled Performance of U.S. RMBS Alt-A Loans Continues To Deteriorate, we reported: “The most disconcerting trend is how quickly the performance of these delinquent borrowers has deteriorated. We continue to see migration from 60-plus-day to 90-plus-day delinquencies within the 2006 vintage, suggesting that homeowners who experience early delinquencies are finding it increasingly difficult to refinance or work out problems, as opposed to being able to ‘cure’ falling behind on payments.” 22 None of these warnings were hidden by S&P and I will gladly provide the Committee with these documents. In addition to these warnings, we also took action in response to subprime deterioration. For example: • On June 1, 2006, almost sixteen months ago, we tightened our criteria through changes in our LEVELS® model targeted to increase the credit enhancement requirements for pools with subprime loans. In announcing these changes to the market, we specifically identified subprime loans, such as “[l]oans with simultaneous second liens (especially those with very low FICO scores)”, as loans “much more likely to default than non-second-lien loans with similar FICO scores.” • Then in February 2007, we took the unprecedented step of placing on CreditWatch negative (and ultimately downgrading) transactions that had closed as recently as 2006. As we informed the market in the accompanying release: “Many of the 2006 transactions may be showing weakness because of origination issues, such as aggressive residential mortgage loan underwriting, first-time home-buyer programs, piggyback second-lien mortgages, speculative borrowing for investor properties, and the concentration of affordability loans.” In a February 16, 2007 Los Angeles Times article, S&P’s announcement was described as “‘a watershed event’ because it means S&P is now actively considering downgrading bonds within their first year.” See S&P to Speed Mortgage Warnings, Los Angeles Times, Feb. 16, 2007. • We continued taking downward action through the Spring. In May we announced that “Standard & Poor’s Ratings Services took 103 rating actions affecting 103 classes of residential mortgage-backed securities (RMBS) transactions backed by subprime, closed-end second-lien, and Alt-A loan collateral originated in 2005 and 2006; we lowered 92 ratings . . . and placed 103 ratings on CreditWatch negative . . . . These rating actions were due to collateral performance.” We also noted that “[m]ost of the transactions affected by CreditWatch placements (and no downgrades) have not experienced significant losses. The placement of our ratings on CreditWatch when a transaction has not experienced significant losses represents a new methodology derived from our normal surveillance practice.” • On June 22, 2007, we announced further ratings actions in an article entitled 133 Subordinate Second-Lien, Subprime Ratings From 2006, 2005-Vintage RMBS On Watch Neg, Cut. We explained that “[t]he downgrades and CreditWatch placements reflect early signs of poor performance of the collateral backing these transactions.” • Then in July of this year, we again took action in response to increasingly bad performance data, including loss levels that continued to exceed historical precedents and our initial expectations. Specifically: 23 • We increased the severity of the surveillance assumptions we use to evaluate the ongoing creditworthiness for RMBS transactions issued during the fourth quarter of 2005 through the fourth quarter of 2006 and downgraded those classes that did not pass our heightened stress test scenario within given time frames. • In addition, we modified our approach for ratings on senior classes in transactions in which subordinate classes have been downgraded. • We also announced that, with respect to transactions closing after July 10, 2007, we would implement changes that would result in greater levels of credit protection for rated transactions and would increase our review of lenders’ fraud-detection capabilities. No one can see the future. The point of these articles and actions, however, is to highlight our reaction to increasing subprime deterioration — looking, as we always do, to historical or paradigm-shifting behaviors to help analyze long-term performance. Consistent with our commitment to transparency we repeatedly informed the market of our view that the credit quality of subprime loans was deteriorating and putting negative pressure on RMBS backed by those loans. And, consistent with our commitment to analytical rigor, we revised our models, took action when we believed action was appropriate, and continue to look for ways to make our analytics as strong as they can be. Impact of The Credit Rating Agency Reform Act of 2006 Earlier this year, the Credit Rating Agency Reform Act of 2006 took effect. As a result, over the past few months, S&P has been actively engaged in the process of implementing the requirements of the Commission’s new Rules regulating NRSROs under the Act. On June 25, 2007 we filed our application to register as an NRSRO. The application includes, among other things, our procedures and methodologies for determining ratings; 24 credit ratings performance measurement statistics; and information related to our ratings analysts and the largest users of our credit ratings. In addition, the application includes a description of our policies for preventing the misuse of material, non-public information and addressing and managing potential conflicts of interest. We also hired a Chief Compliance Officer who is responsible for administering and overseeing these policies and procedures and ensuring compliance with applicable securities laws. Additionally, S&P has continued its ongoing efforts to develop and streamline internal record-keeping policies and procedures in order both to ensure the integrity of the ratings process and to satisfy Commission requirements that records be available for inspection. We recently received a notice of examination from the Commission seeking the production of a substantial amount of documents that may relate to the issue of the potential conflict of interest discussed above. We are in the process of complying with this notice. S&P supported final passage of the Credit Rating Agency Reform Act and remains committed to that Act’s stated goal of improving ratings quality for the protection of investors and fostering oversight, transparency and competition in the credit rating industry. Given that we are relatively early in the process of seeing this new law fully implemented, we would respectfully urge you to allow the Commission to proceed with its task of enforcing the provisions of the new law and the regulations so recently adopted before Congress proposes any further actions. Conclusion I thank you for the opportunity to participate in this hearing. Over the past several decades, S&P’s consistent approach has been to evolve our analytics, criteria, and review 25 processes when appropriate, and you can expect that same approach in light of new consumer credit behaviors in all markets, including residential mortgages. Let me also assure you again of our commitment to analytical excellence and our desire to continue to work with the Committee as it explores developments effecting the subprime market. I would be happy to answer any questions you may have. 26