Since the revolution in modeling residential-mortgage risk in the U.S. began in late 1995, proprietary credit and mortgage scoring systems have become pervasive in global mortgage products, with major originators fusing such technology into their underwriting processes. In 1998, when Standard & Poor's introduced its first version of a modeling software known as LEVELS(tm), only 50% of the prime mortgages submitted for ratings included a credit score on the tape and less than 30% of the subprime mortgages incorporated a credit score in their underwriting data file. By the end of the second quarter of 2002, virtually 100% of the newly originated mortgages submitted for ratings incorporated credit scores, and in some cases mortgage scores. All major mortgage underwriters in the U.S. blend some form of credit scoring and/or mortgage scoring algorithm in the process of qualifying borrowers and identifying the appropriate mortgage products by borrower characteristics and credit experience. This rapid and extensive growth in scoring has had a significant impact on the U.S. mortgage-backed securities market. The global trend appears to support similar developments by the local financial institutions and U.S.-based companies.
Refining Mortgage Risk Categories
The use of credit scores and mortgage scoring models in originating mortgages has led to a significant refinement in the categorization of risk. The differentiation of borrower risk in credits by A' quality, A-', B', etc. has in many cases been replaced by score "buckets," which allow a significant refinement in the allocation of risk to various types of securitized transactions. The ability to sort risk by score has resulted in a reduction in subordination requirements for both prime and subprime pools over the past six years. In the prime market, the AAA' subordination requirement has gone from an average of 6.25% in 1996 to 3.31% at the end of 2001 (see table). This reduction in enhancement requirements has not been lost on originators and conduits, which are using models to sort the loans according to risk categories and scores to maximize their securitization proceeds. In addition, models have supported improved risk-based pricing at the point of sale. In the subprime arena the ability to adequately price risk with the appropriate coupon has led to lower overall enhancement levels because of the adequacy of the excess spread that is provided through a more refined risk-based pricing process.
The outcome of Standard & Poor's reviews of several global proprietary systems is that lower enhancement requirements may be available to issuers of mortgage-backed securities in the global markets. These benefits occur when Standard & Poor's is provided with adequate information from the proprietary scoring engine along with loan-level performance data that allows it to map the proprietary scores to the Standard & Poor's risk grades utilized in the U.S. models.
This mapping has allowed Standard & Poor's to calibrate issuer-specific model scores at the loan and pool levels and to incorporate the benefits of refined loan grading and stratification in the sizing of credit enhancement requirements at each ratings level.
The ability to map proprietary credit and mortgage scores to the Standard & Poor's grading system requires that issuers provide loan-level information on incorporating scores for the loan and the resulting loan and pool performance measures. Standard & Poor's can then modify the ratings methodology embedded in the U.S. model to develop an issuer-specific model derived from the proprietary scores and loan-level information provided by the specific arranger or issuer. Any improvements in the ability to quantify the risk at the securitization level and the investor reporting level can have a significant impact on the economics of the transaction as well as the acceptance of the resulting securities by local and international investors.
Proprietary credit and mortgage scoring models have enjoyed remarkable success and continued development in the U.S., and the applications that are under development and incorporated in the global arenas are as good, and in some cases better, as predictors of risk than those developed in the U.S. Standard & Poor's expects this focus on modeling to expand the proprietary development efforts and stimulate the creation of national databases that will support the generation of credit scores that should make the origination and securitization of residential mortgages that much more efficient.
S&P's DACSS System
As automated and electronic property appraisals continue to revolutionize the lending industry - offering high-tech, expeditious and low-cost alternative appraisal methods for real estate transactions - it has become increasingly critical for originators to analyze a borrower's credit risk at point-of-sale, especially for loans and properties that may be assessed by alternative means.
By doing so, lenders can more strategically manage through the life of the loan, from the efficient analysis of appraisal requirements at origination, up to the competitive pricing of the loans during securitization.
This type of holistic, strategically focused view of the lending business is certainly where the future of residential-mortgage technology is headed, and Standard & Poor's revolutionary analytical software tool, DACSS (Documentation and Collateral Scoring System), launched in January of 2002, is the first platform to fulfill the entire panoply of originator needs. Designed specifically for residential mortgage loans targeted at the non-conforming market, DACSS is an invaluable tool that evaluates the attributes of a U.S. mortgage loan application and the credit quality of the borrower.
But more importantly, it assigns alternative document and collateral requirements at point-of-sale in accordance with Standard & Poor's ratings criteria - without assigning a corresponding increase to the foreclosure frequency or loss severity expectations.
The advantage of this? If an issuer knows that a loan is "DACSS-eligible", or S&P-approved, at point-of-sale, that loan will not be charged additional loss coverage when it is sold into the secondary mortgage market. Even though these loans are eligible for reduced or streamlined documentation, investors and issuers will know in advance which loans meet S&P's ratings criteria, giving them an unparalleled pricing advantage on whole-loan bids.
Moreover, if a loan is pre-approved by DACSS for the use of an automated valuation model (AVM), a borrower will only be asked to pay a minimal fee for this electronic appraisal, as opposed to the far more expensive traditional appraisal process. Ultimately, this can only increase a lender's volume and enhance its customer service capability - a win-win situation for both the lender and the borrower.
DACSS, therefore, is the quintessential full-service securitization tool for the lender: at the front-end, it determines reduced income documentation and appraisal requirements for a first-lien non-conforming loan, including a determination of which AVM systems the loan is eligible for; at the back-end, it is an effective analytic tool for bulk loan sale/acquisition valuation, ensuring competitive risk-based loan pricing for a securitization.
It is also ahead of the curve in its technology. With the advent and popularity of AVMs over the last several years - giving lenders and borrowers an accurate, quick and cost-effective means of computing appraisal values - originators have been searching for a means of efficiently determining which appraisal method is appropriate for a particular borrower. DACSS will approve a loan at point-of-sale based on appraisal type, assuming other risk factors are met, and recommend which appraisal method the loan is eligible for, eliminating much of the guesswork for the lender.
Perhaps most significant, however, is the prominent place that DACSS takes in the evolution of the U.S. non-conforming residential mortgage market, a growing sector that covers loans in excess of $300,700.
In the past, there has been little standardization or guidance in the non-conforming market in terms of which loans are eligible for alternative treatment. In contrast to the conforming market, non-conforming loans, historically, would automatically require the full traditional documentation and appraisal process for both income and property, from the get-go - there would be no flexibility for those requirements. With the use of DACSS, however, a non-conforming loan pre-approved by the system for streamlined or alternative documentation and appraisal will not get hit with a loss-coverage penalty later on for bearing this alternative assessment.
And this benefit is a real breakthrough for the non-conforming secondary capital markets. When an issuer sells a loan to an investor, the calculation of credit enhancement or support for the loan is determined by many factors, including the credit quality of the borrower, as well as mortgage-loan characteristics, such as how the properties were appraised. For a DACSS-approved alternative loan, however, the prior knowledge of S&P acceptability accelerates the entire process, leading to reduced due diligence expenses and an unmatched competitive edge for both the investor and the issuer, who can target their efforts more carefully.
Furthermore, DACSS's immense capabilities have arrived in the residential mortgage market at the right time, analysts say. As interest rates remain low and refinancing volume declines, originators continue to look for efficient methods to increase volume. And the need is there for the growing non-conforming sector: in the first quarter of 2002 alone, some $7.01 billion of the collateral backing Alt-A deals were non-conforming loans, or 65% of total issuance.
DACSS, therefore, reflects the future of the mortgage origination and securitization business: it is a one-stop shop for loan-level income and collateral documentation recommendations and analysis, as well as an effective strategic tool for secondary market pricing.