© 2024 Arizent. All rights reserved.

Computer Model Estimates Prepayment Forecasts

Dynamic prepayment forecasts, those that change with the market, have been increasing in demand over the years. The Mortgage Industry Advisory Corp. (MIAC) has developed a program that can predict prepayment medians on a daily basis.

Because Wall Street dealers typically only provide prepayment models on the first and 15th of each month, if a dramatic change occurs in the market between models, traders in the industry would traditionally have to wait until the model is released to take that into account. The system, called Mortgage Industry Medians (MIMs), can estimate prepayment models on a daily basis.

"You tend to have some inconsistencies, some noise basically going on," for those dealers that only issue models twice a month, compared with others which may update weekly, explained Bob Husted, principal at MIAC. "Mortgage Industry Medians is really a predicted model, to predict what dealer consensus, prepayment speeds will be in the intervening periods where an update has not been made."

Husted added that while this model will not affect "plain vanilla" mortgage-backed securities, it will affect structured products and mortgage servicing rights. "Getting good price discovery is really going to be dependent on getting good prepayment forecasts in those periods, whereas typically on the hedges there's much greater price discovery about hedges," he said.

The MIAC's goal was to provide, in the advent of Federal Accounting Standard 133, set to go into effect in June, a correlated regression model to estimate prepayment speeds during the interim periods of dealer published models. "Under FAS 133, anybody that might be hedging mortgage-related assets, there are greater potential penalties in terms of PNL impact if they fail so-called hedge effectiveness," said Husted. "The usefulness of MIMs, while very much existent today, will increasingly be so under FAS 133."

The MIAC is using Bond Market Association data, which consists of 11 contributing dealers, the largest sample of prepayment medians. MIMs takes into account a given interest-rate environment on a particular date to determine its "highly predictive" estimate, which will be published on its Web site (www.servicing.com/miac) and updated daily.

Not A Replacement

Husted noted, however, that while "there's nothing perfect short of an actual update and a dealer forecast, if you want to come up with a very highly predictive model of what the forecast would be when those updates are not otherwise available, MIMs is really what would get you there."

He added that MIMs is not intended to be a replacement to dealer forecasts, but to act as a supplement to them. "In a perfect world the dealers would update their forecasts daily," said Husted. "MIMs is basically filling a gap."

MIMs is not a prepayment model, he said, but "it's just a way to estimate dealer consensus speeds in different interest rate environments."

Prepayment experts, while not familiar with the exact mechanics of MIMs, do believe that more frequent models are necessary. "I've heard customers mention the lag time between updates on the medians." said Dale Westhoff, managing director of MBS at Bear, Stearns & Co. "I know a lot of customers have full access to dealer projections every day, so it's not like they're not updated in between. But to the extent that they're using just the medians to perform analysis, it certainly would be helpful, if it's accurate."

Andrew Davidson, president of Andrew Davidson & Co., a prepayment analytics company, offered similar remarks. "Hedging is a dynamic activity and you can't be using stale forecasts in evaluating hedging and marked market activity. I'm definitely in favor of that direction," he said. "The medians as they stand now, while they're valuable at understanding what different dealers' views of prepayments are, they're not necessarily that valuable for valuing portfolios intra-month, because they're really only valid as of the date of the forecasts."

The MIAC has been testing the model since October, and sent MIMs through rigorous testing, such as feeding in prepayment information over the past year into the system to see how accurate the estimate was to the actual market events. The system is now ready to be fully executed.

MIMs' Other Use

Husted noted that besides determining current day prepayment estimates, MIMs can also be used to predict how prepayments would respond to future changes in the market. "If you're stressing your analysis, if you're running hedges and you want to test hedge effectiveness under different conditions that would not be available in the current forecast method, MIMs would be very effective as a forecasting tool," he said.

So far, MIMs has received a great deal of support in the industry, including customers as well as regulators.

For reprint and licensing requests for this article, click here.
MORE FROM ASSET SECURITIZATION REPORT