Partnership Aims to Help Lenders Weigh Short-Term Prepayment Risk

Lenders and analytic companies have developed a vast body of data about prepayment rates over the years, but the sustained drop in interest rates that shaped the lending environment over the past three years challenged even the best prepayment predictors.

Specifically, few anticipated how fast many new loans would prepay. Some were gone from the books within a few months of being boarded on the system. Now, two firms are teaming up to offer lenders a new tool for judging short-term prepayment propensity.

Applied Financial Technology here has entered into an agreement with McDash Analytics to provide the nation's leading lenders and mortgage financing companies with short-term prepayment projections as well as housing turnover and refinancing scores.

The strategic partnership combines the scoring expertise of AFT with the large database of performance information involving nearly 20 million mortgages, or about one-half of the mortgage market, that is owned by McDash Analytics.

The companies say that the partnership will give lenders tools to help them enhance the accuracy of prepayment projections and allow greater insight into mortgage values and future behavior.

The data from McDash do not include borrower-specific information, and users can only identify their own loans. The rest are anonymous as far as individual lenders are concerned.

Graham Williams, chief operating officer at AFT, says the technology gives users more "granularity" to distinguish prepayment behavior among similar loans.

He said the AFT prepayment score is a loan level modifier to standard prepayment models. Higher scores indicate a higher propensity to prepay relative to loans with the same coupon, maturity and loan type over long periods of time and all interest rate environments.

AFT provides prepayment analytics to large mortgage servicers, broker/dealers, insurance companies, hedge funds and other parties that have a substantial and significant exposure to prepayment risk, Mr. Williams said.

"From our perspective, having access to the loan level information that is embedded in the McDash database allows us to better understand the key drivers of prepayments," Mr. Williams said.

Historically, prepayment models have looked at key statistics such as the coupon rate, the loan type and the mortgage term. But other loan level data, such as geography, loan purpose, loan-to-value ratio and credit scores, can help to refine projections.

"The prepayment behavior of various combinations of those things are different. And historically, the market hasn't been able to understand that," he said.

Loans that previously would have been treated as identical can be differentiated for a more detailed analysis of prepayment behavior.

"We view this to be fairly unique and new. We feel that over time these scores will be used in buying and selling mortgages much like a FICO score is used today," Mr. Williams said.

Ted Jadlos, chairman of McDash Analytics, said the company's loan data includes 15 years of history.

"The most important thing when it comes to data mining tools," he told MSN recently, "is that the more granular the data, the more easy it is to make apple-to-apple comparisons."

Essentially, the AFT scores and measure of short-term prepayment propensity gives users a "mortality forecast" when they are evaluating a portfolio of loans or servicing rights. That's particularly important given the failure of many prepayment models to anticipate how quickly loans would run off during the peak of the refinancing boom. "The short-term forecasts just haven't fit the short-term reality of prepayment speeds," Mr. Jadlos said.

Copyright 2004 Thomson Media Inc. All Rights Reserved. http://www.thomsonmedia.com http://www.mortgageservicingnews.com