DBRS Updates its Assumptions for Monitoring Ratings on U.S. RMBS

2010 comprised another year of challenges in the U.S. mortgage market. Unemployment rates appeared to have stabilized but at levels significantly higher than normal. Housing prices remained soft, with supply exceeding demand, leaving many homeowners underwater in their homes and distressed borrowers with limited refinancing or sale options. Programs designed to modify loans and relieve the foreclosure pipelines have been moderately successful. Consequently, defaults and loss severities on liquidated properties have continued to remain high.

The impact of the aforementioned factors on Residential Mortgage-Backed Securities (RMBS) and ReREMIC transaction performance has been substantial and unprecedented. Therefore, in response to observations made in its routine examinations of market conditions and transaction performance throughout the RMBS sector, DBRS has made adjustments to the assumptions it uses to monitor U.S. RMBS and ReREMIC transactions*. The updates relate to loss severity, default timing curves and the number of cash flow stress scenarios run, which are based on a combination of prepayment speeds, interest rates and loss timing.

DBRS is in the process of reviewing all of its outstanding ratings in the U.S. RMBS and ReREMIC sectors, utilizing the updated assumptions and additional cash flow scenarios described below. The review will be conducted in phases and DBRS will publish the results of each rating committee as soon as possible.

Loss Severity
DBRS reviewed actual loss severity data on substantially all existing transactions in Intex with a concentration on the most relevant and recent vintages from 2004 to 2008. As a result, DBRS has increased its base case loss severity assumptions by sector and vintage, as indicated in the table below.

Loss Severity (Updated):
Vintage−−Subprime−−ALT_A−−Prime−−HELOC−−Option ARM−−2nd Lien
2008−−−−−−70%−−−−−55%−−−50%−−−100%−−−−−60%−−−−−−100%
2007−−−−−−75%−−−−−60%−−−55%−−−100%−−−−−65%−−−−−−100%
2006−−−−−−75%−−−−−60%−−−50%−−−100%−−−−−65%−−−−−−100%
2005−−−−−−70%−−−−−55%−−−45%−−−100%−−−−−60%−−−−−−100%
2004−−−−−−65%−−−−−55%−−−45%−−−100%−−−−−55%−−−−−−100%
2003−−−−−−60%−−−−−50%−−−40%−−−100%−−−−−55%−−−−−−100%
<=2002&#8722;&#8722;&#8722;&#8722;55%&#8722;&#8722;&#8722;&#8722;&#8722;45%&#8722;&#8722;&#8722;35%&#8722;&#8722;&#8722;100%&#8722;&#8722;&#8722;&#8722;&#8722;55%&#8722;&#8722;&#8722;&#8722;&#8722;&#8722;100%
2nd Lien:&#8722;&#8722;100%&#8722;&#8722;&#8722;&#8722;&#8722;100%&#8722;&#8722;100%&#8722;&#8722;100%&#8722;&#8722;&#8722;&#8722;&#8722;100%&#8722;&#8722;&#8722;&#8722;&#8722;100%

Additionally, to the extent a specific transaction has experienced loss severities that are significantly higher than the baseline numbers in the above table, DBRS takes the actual loss severities into consideration when monitoring the rating.

Default Timing Curves
Historically, defaults generally peaked between years two and four of a loan, and after year five were rare because borrowers either had built up enough equity due to home price appreciation or were able to refinance/prepay. In the current environment, this is not the case. Additionally, DBRS has observed that a) delinquencies and defaults continue to rise as RMBS transactions age and b) current non-delinquent loans are under tremendous stress due to significant home price declines and high unemployment rates. Even the very seasoned vintages (pre-2005) that were considered good performers are now exhibiting considerable default activity in the past twelve to 24 months. This is further exacerbated by slow prepayment speeds which prolong the life of deals and expose them to greater potential for defaults. As a result, DBRS has adjusted its default timing curves to assume additional defaults occur, even if the historically-proven default curves have ended.

The adjustment made depends on the seasoning and pool factor of the transaction. For example, 2007 vintages can generally expect another 30% of defaults to occur in the future in addition to the existing delinquency pipeline. Similarly, pre-2007 vintages are projected to experience an additional 20% in defaults.

Cash Flow Scenarios
The adequacy of credit enhancement to protect investors is also a function of each transaction’s prepayment speed, interest rates and the timing of losses. Given the complexity of RMBS waterfalls and how sensitive bonds are to cash flow assumptions, DBRS believes it is prudent to approximate various combinations of stressful scenarios in order to adequately test the resilience of the bonds. Accordingly, DBRS runs a total of 40 cash flow scenarios for each transaction. The scenarios include five prepayment speeds from 5% to 25% constant prepayment rates (CPR) at 5% increments under two prepayment conventions to equal 10 prepayment scenarios (5 x 2 = 10). For each of the 10 prepayment scenarios, both a front-loaded and a back-loaded loss timing curve are applied, bringing the number of scenarios run to 20 (10 x 2 = 20). Finally, for each of these 20 scenarios, an upward and downward interest rate stress is applied resulting in a total of 40 scenarios run (20 x 2 = 40). The most conservative outcome of the cash flow scenarios is used in deriving a rating for each security.

Notes:
DBRS rating definitions and the terms of use of such ratings are available at www.dbrs.com.

*The applicable methodology is U.S. RMBS Surveillance dated April 2009, which can be found on our website under Methodologies.