Ratings

We published guidelines for increasing overcollateralization (O/C) floor levels, calculated on a deal-specific basis, in many U.S. RMBS transactions to provide additional credit support. We believe these higher O/C floor levels (along with alternative trigger, step-down, and principal payment mechanisms) will better enhance the stability of our ratings in U.S. RMBS transactions. Investors have expressed concern that excess spread has been released from deals prematurely, allowing payment to net interest margin security (NIMS), equity, or residual noteholders at the expense of preserving credit support for the more senior bonds.

In keeping with our aim of greater transparency and information, we are collecting more data concerning the procedures issuers and originators use in order to assess the accuracy and integrity of their data as well as the measures they take to detect possible fraud. We believe this will allow us to better delineate the operational capabilities of originators and conduits and their ability to deliver quality loans and data.

To that end, we have enhanced our approach to determine the operational quality of mortgage originators that securitize loans in the RMBS we rate. Our criteria focus on originators' core functional capacities having both a direct and indirect impact on the quality and accuracy of loan-level risk metrics, and apply a holistic analysis to the overall quality of a given originator. As such, we expect that the implementation of these criteria and the review and ranking of mortgage originators should afford us a deeper insight into the quality of data in loan pools submitted for analysis. We believe this will enhance our ability to develop valuable public opinions on originators and issuers of RMBS. We will review and rank originators according to the operational quality of eight functions, each of which we consider relevant to the operational quality of mortgage originators:

• Management and organization;
• Risk management;
• Underwriting;
• Regulatory compliance;
• Appraisal management;
• Broker/correspondent/retail loan officer management;
• Prefunding data quality control; and
• Postfunding quality control.

In the article, "S&P Focuses On Appraisals For Residential Properties And Encourages Use Of Automated Valuation Methods," published Jan. 22, 2008, on RatingsDirect, we discuss our belief that greater transparency and improved data quality are obtained if issuers use an automated valuation method (AVM) on all securitized loans as a supplement to traditional appraisal methods and to make those values available to both us and to RMBS investors. AVMs are statistically based computer programs that use real estate information, such as comparable sales, property characteristics, tax assessments, and price trends, to provide an estimate of a specific property's value.

It is also our opinion that the appropriate use of AVMs can diminish the risk associated with the most commonly reported fraudulent appraisal practices by offering a second opinion, outside of the origination process, on the collateral's value. This supplemental data will be used as part of our operational risk analysis, as it will facilitate our ongoing efforts to enhance our analytics and processes and help us to provide additional insightful research to the mortgage and securitization markets.

Another measure we will be taking to increase transparency is to publish "what-if" scenario analyses that explain our key rating assumptions and the potential effect on outstanding ratings if these alternative economic scenarios were to occur. The composition and factors we analyze will vary, but we intend to focus on determining what level of losses 'AAA' and 'BBB' tranches would be able to withstand given three different economic scenarios for each unexpected event. This information will give investors more information about our ratings and the assumptions used to generate those ratings, aiding investors in their efforts to better understand the effect these economic what-ifs will have on our potential rating actions.

We will soon begin publishing U.S. RMBS presale reports with the aim of offering investors and market participants additional information that supplements our traditional ratings.

Our detailed presale reports may include information such as:

• Collateral characteristics;
• Significant geographic concentration;
• Mortgage insurance details;
• Comparisons to prior transactions (by issuer and/or originator);
• The impact of different scenarios within our cash flow modeling, such as varying interest rate paths; and
• Structural elements including O/C floors, triggers, credit enhancement multiples, and the impact of derivatives.

Beginning with transactions that close on or after May 1, 2008, Standard & Poor's has requested that issuers of U.S. RMBS send us monthly performance data customarily sent to trustees and other third parties, and include representations and warranties that all loan and performance data fields sent to us are true and correct at the time we assign our rating; we would like such information to be provided to us monthly for the life of the transaction. We are also working closely with the American Securitization Forum and other market participants to create a uniform reporting package of loan-level data to be available to all market participants both at the time of new issuance and ongoing for monitoring transactions in the secondary market.

With this monthly flow of loan and performance data, we plan to develop a loan-level approach to the surveillance of outstanding U.S. RMBS ratings and move forward with efforts to conduct a more granular analysis of the factors influencing borrower behavior in the residential mortgage market. This additional data may also help identify patterns and trends in the market, enhance our surveillance of outstanding transactions, and allow us to provide additional insightful research to the mortgage and securitization markets. The regular transmission of loan-level performance data to Standard & Poor's for rated U.S. RMBS transactions expands upon our existing practice in Europe and Australia.

We recently published a glossary of terms associated with loan variables and performance fields used at new issuance and for surveillance. We are working with industry participants to standardize terms and calculations, which should greatly improve transparency and data quality. Originators and issuers specifically requested such a glossary (see "Glossary Of Terms For Requested Additional Loan-Level Fields For New U.S. RMBS Ratings," published April 2, 2008, on RatingsDirect.)

We recently revised some assumptions in the latest release of our U.S. mortgage analytic model, LEVELS® 6.3. The revisions change the credit enhancement assumptions for first-lien, Alternative-A, and subprime transactions. The changes affected all deals closing on or after March 14, 2008.

The LEVELS 6.3 model revises the loan-level assumptions that determine the probability of default given relationships among various loan and borrower characteristics, including FICO scores and combined loan-to-value ratios. Consequently, LEVELS 6.3 incorporates these new assumptions based on recent loan performance into loan-level criteria. Changes in the LEVELS 6.3 release include:

• The ability to analyze home equity line of credit loans;
• Loan-level delinquency adjustments;
• A Massachusetts foreclosure timeline extension;
• Updated rating levels of mortgage insurers; and
• Updates to Standard & Poor's House Price Index with recent Office of Federal Housing Enterprise Oversight data.

We are evaluating the feasibility of using, in future releases of the LEVELS model, stochastic modeling techniques to help account for alternative macroeconomic factors and to add further stress scenarios to our default and loss modeling. There would be some advantages in using such techniques:

• The model may forecast macroeconomic factors for up to 100,000 different scenario simulations;
• Stresses may occur at different rating levels in a given 30-year period, which is significant now that triggers will be modeled in Standard & Poor's Interest Rate Evaluator (SPIRE™);
• Loan behavior in specific economic situations may be captured and used to mimic behavior in similar environments;
• Each macroeconomic factor may be forecasted in correlation with other macroeconomic factors; and
• Each month's prediction may be based on the previous month's data as well as other relevant macroeconomic factors seeking to achieve realistic scenarios.

Anticipated results may include:

• Rating levels tied to confidence intervals;
• The performance in each of the 100,000 scenarios captured and reported; and
• The addition of further information on the effect that varying economic scenarios will have on our outstanding ratings.

SPIRE V 3.1 was released April 8, 2008, and immediately put into effect for all of our rated RMBS transactions. New features include:

• Loss/delinquency triggers and the ability to alter principal distribution amounts (multiple waterfalls);
• Stress scenarios that include the use of multiple paths and assumptions;
• The calculation of a transaction-specific O/C floor;
• New default curves;
• A credit enhancement multiple; and
• Standard & Poor's assumptions on step-down, delinquency, and cumulative loss triggers.