When MBS issued by Freddie Mac, Fannie Mae and Ginnie Mae dominated the secondary market, these government-sponsored enterprises assumed virtually all of the credit risk, leaving investors to worry principally about prepayment risk. Over the last two to three years, the non-agency share of the secondary market has exploded, leaving investors to confront the "two-headed risk monster" of both prepayment risk and credit risk. To account for the addition of credit risk, Andrew Davidson & Co. has developed a behavioral model that provides projections of prepayment, delinquency, and default for non-agency MBS.
Kyle Lundstedt, the director of credit modeling at AD&Co., explained that credit risk traditionally has been the focus in origination or underwriting, while prepayment risk was the domain of portfolio management. "Over the last two decades, agencies comprised the vast majority of available mortgage securities, so investors principally focused on interest rate and prepayment risk," Lundstedt said. "Now, over half the secondary market is comprised of non-agency securities with a wide range of collateral characteristics, so investors must assume the credit risk themselves."
Lundstedt emphasized that the prepayment and credit risks are fungible and completely interdependent, an issue of primary importance in the subprime credit sector. "Every month, there are some subprime borrowers in a position either to prepay or default. Consider a subprime borrower who has made 11 consecutive payments, yet whose house value has dropped slightly below the current mortgage amount. Making the twelfth payment puts them in a position to refinance; however, the house price decline might push them to default. The prepayment and credit risks are interrelated."
Consequently, AD&Co is introducing a credit model to provide investors with a tool for evaluating these risks. The model uses many of the same loan-level variables as those used in the firm's loan-level prepayment models. In effect, the prepayments variables in the credit model - the refinancing spread, loan-to-value ratio, FICO score, etc. - are essentially the same as in the firm's widely used prepayment models. Moreover, AD&Co is using the same loan-level information to project the probability of delinquency and default, as well as the severity of loss.
"The key thing that distinguishes the credit model is not the data used, but the fact that we incorporate the effect of the loan's payment status on borrower behavior," Lundstedt explained. It is generally accepted that, if a loan's payment status is current, termination of that loan constitutes a prepayment. Conversely, termination of a foreclosed loan is almost certainly a default. It is not clear, however, how to treat loans that pay off from, say, 60 days past due, Lundstedt added. He also noted that investors, in addition to termination from prepayment and default, also care about delinquency triggers and their corresponding effect on bond cash flows.
The new AD&Co credit model covers most credit sectors within the non-agency market, including jumbo, Alt-A, low FICO and high LTV loans. "Traditionally, investors had to identify the credit sector occupied by a pool of loans in order to run the correct behavioral model," Lundstedt said. "One of the real innovations of the AD&Co. model is that credit sector is not a variable; loans are evaluated based upon observable characteristics, not loosely defined marketing terms like "subprime" or "Alt-A".
Lundstedt said that this new credit model has a unique place in the market. For several years, rating agency tools such as Standard & Poor's Levels or Moody's Investors Service's Mortgage Metrics have met the needs of MBS issuers interested in the rating agencies' perspective on the credit risk of their prospective bond issues. "This is the first credit model that is targeted specifically to the needs of non-agency investors," Lundstedt said. Such investors typically include hedge funds, insurance companies, REITs and money managers.
Aside from its new credit model, AD&Co. has built additional tools critical to evaluating MBS credit risk. In particular, they have created simulation software for house price indices, similar to that used for Monte Carlo-based OAS modeling. House prices have become an increasingly important economic risk driver for both prepayments and defaults. Lundstedt noted that investors already are familiar with home price appreciation effects on "cash out" refinancing, which affect prepayment risk even when interest rates remain constant. "Simulation of house price indices," he said, "enables investors to understand both prepayment and default behavior under a range of future house price scenarios." AD&Co. will be working to integrate its new tools into various software systems such as Wall Street Analytics and QRM. "Investors hopefully will find our new tools substantially increase their ability to understand the credit risk of MBS," Lundstedt said.
(c) 2006 Asset Securitization Report and SourceMedia, Inc. All Rights Reserved.