Residential mortgage-backed securities (RMBS) with a high concentration of California mortgages issued since the mid-1990s are not showing the high mortgage losses that once plagued similar securities from the early nineties. When California began its recession in 1990, housing prices were soaring, and the California economy was poised for disaster as layoffs increased across the state. The overheated housing markets started sliding shortly thereafter as home prices decreased and borrowers defaulted. Losses accumulated on deals collateralized primarily by California properties, and bondholders of subordinate classes felt the brunt of mortgage losses. Because of the poor performance of these securities from the early 1990s, deals collateralized by pools with significant concentrations (30% or more) of properties located in a single state, usually California, have carried a stigma that continues today. However, over the past decade, originators, issuers, and rating agencies improved their processes, alleviating some of the risk of geographically concentrated mortgage pools.
Today, originators benefit from the use of automated underwriting systems (AUs). Implementation of AUs has streamlined origination practices, creating efficiencies and enabling lenders to originate with more consistent standards. Borrowers meeting the originators' guidelines are approved more quickly, and the risk profiles of these mortgagors are more consistent and predictable. Credit and mortgage scores give additional insight into identifying the relative risk of borrowers and appropriate risk-based pricing, and automated valuation models provide a less costly and less time-consuming alternative for evaluating properties and are often more reliable, with less biased results.
As borrower and mortgage evaluation has evolved, so has the overall process of securitization. Segregation of mortgage loans into distinct sectors (prime, alternative, and subprime) enables lenders to extend credit to larger and more varied classes of borrowers, issuers to issue securities with distinct borrower-credit-risk profiles, and investors to purchase securities that more certainly satisfy their risk appetites. The result of these advances in the origination and securitization processes is more performance that is predictable for all RMBS, even those with high concentrations in a single state.
The sustained weakness in California's housing markets in the early 1990s would lead one to expect that, on average, deals from this period with higher levels of California concentration would suffer more losses while those with moderate levels would suffer fewer losses. One way to illustrate the improvements that have occurred is to examine loss levels of deals with high California concentrations from the early nineties and those securitized more recently. In order to undertake such an analysis, it is necessary to derive a representation of the deals that existed in the early nineties. One informative and practical means is to limit the selection of securities to those that are composed entirely of loans from California. This offers the advantage of comparing deals perceived to have the highest level of risk, while allowing one to extrapolate about deals with relatively more moderate California concentrations, say 50% and higher. The Fitch Ratings surveillance database includes more than 20 deals issued from 19911993 backed entirely by California properties. Mortgage loans from these deals were originated between 1990 and 1993. The worst performing deal has upwards of 11% losses of its original collateral balance, while the best lost less than 1%. All the deals were "nonconforming, A quality," which connotes a mixture of prime and alternative loans. It is estimated that prime loans accounted for approximately 80% and alternative loans 20% of the total. From these deals, a subset representing the average population was taken, and an aggregate loss curve was derived. The aggregate loss curve begins to plateau at roughly 60 months of seasoning, with losses reaching an average range between 2%3%.
This aggregate loss curve provides a benchmark with which to compare deals more recently issued. Mortgage pools with loss curves that resemble that of the early nineties represent generally riskier pools, while lower, flatter loss curves represent less risky, more recent pools. Of the deals originated since 1996, only a handful had 100% California concentrations. All these deals have performed better than the benchmark as they have seasoned (see chart on p.12). Not surprisingly, deals with smaller California concentrations have also outperformed the benchmark.
The question remains of whether there are any other deals that have performed like the benchmark. There are, but only within the subprime sector. One interesting example of a mortgage-backed security with both a high concentration of California properties and losses that are very reasonable for subprime is New Century Home Equity Loan Trust 1997-NC5. At the date of issuance, more than 50% of the properties collateralizing this security were located in California. While approximately two-thirds of the mortgagors were considered A+ or A, denoting they met relatively stringent criteria with respect to mortgage payment history (maximum of one 30-day late payment during the past year), the remaining one-third of the mortgagors experienced more significant credit problems during the past 12 months.
This pool, with its many California properties and mortgagors with tarnished credit histories, suffered larger losses than did the benchmark pool. However, unlike the losses suffered by the benchmark pool, the mortgage losses of the New Century 1997-NC5 pool are neither surprising nor alarming. First, because not all these mortgagors have pristine credit histories, we expect higher delinquency and loss rates. Furthermore, because California housing markets have been stronger since 1996, the loss suffered on every post-1995 liquidated property was not compounded by extraordinarily harsh conditions in the housing markets. Finally, losses are not alarming because the lowest rated class itself has adequate credit enhancement (in the form of subordination, overcollateralization, and monthly excess spread) to protect bondholders. This security was structured to accommodate gross mortgage losses of 4.5% at the BBB' rating category. Through February 2002, despite mortgage losses, New Century 1997-NC5, class B (rated BBB'), realized no certificate losses. This security - characterized by mortgage losses, a risky borrower profile, and a high California concentration - has outperformed its early 1990s prime predecessors.
Because of changes in origination and securitization, it is unlikely that RMBS with significant exposure to the California housing markets today would experience losses similar to those suffered in the early 1990s. Better risk assessment of both borrowers and securities has enabled the RMBS market to develop in a positive direction. Although the exposure to California once was a source of concern, developments within the market have compensated for the concentration risk. In fact, the improvement in California economic conditions in the late 1990s (and associated RMBS credit performance) is greater than that of the U.S. market overall. The similarity in behavior between a 1997 security collateralized by California mortgages made to subprime borrowers and the early 1990s California "nonconforming, A-quality" pools (reflecting approximately an 80/20 mix of prime jumbo and alternative-A loans) demonstrates that weaker borrowers and stronger collateral can better support a mortgage-backed security than stronger borrowers with weaker collateral.
While "100% California" may no longer be frightening, mortgage-backed securities collateralized by properties exclusively or predominately located in "hot" markets remain of concern. Home prices can increase at a faster rate than is economically sustainable, creating speculative bubbles in the housing markets. Appreciating markets support the securitization process by enabling defaulting borrowers to sell their properties, avoiding foreclosure, and allowing servicers to liquidate foreclosed properties without loss. However, housing markets that appreciate too rapidly can deflate as quickly, exacerbating the potential for loss upon liquidation. Santa Clara County experienced an increase of almost 36% in its overall housing market during 1999, compared with increases of about 10% in Los Angeles County and less than 10% in the country overall. Although the projected housing price index for Santa Clara County anticipated a downturn of about 20% in 2001, the downturn did not occur. To the extent that hot metro areas predominate mortgage-backed security pools, investors should remain cautious. However, with sufficient geographic dispersion between north and south, "100% California" no longer represents the concentrated geographic risk that it did in the early 1990s.
