Geography Affects Prepayment Risk
With rates once again slipping in the third quarter, lenders - just recovering from the biggest refinancing boom in history - have seen prepayment rates rise again.
But in the mortgage secondary market, people are still failing to accurately price loans based on their propensity to refinance, according to a prepayment expert here. In part, that's because much of the mortgage investor and servicing marketplace lacks an understanding of the geographic variable behind a borrower's propensity to refinance.
Michael Bykhovsky, president and CEO of Applied Financial Technology, noted that some things have changed since the mid to late 1990s, which is the period of time on which many current models rely to make prepayment projection for the current housing market environment. For instance, existing home sales now run at a rate of about 6.2 million units per year, up from 4.5 million in 1996, with a very large increase even accounting for the increase in the housing stock.
"Just from that you would expect a higher rate of prepayments related to housing turnover than would be implied by the history - most models underestimate that," Mr. Bykhovsky said.
He also believes that the influences of geographic factors that are partially linked to the home price appreciation index are more nuanced than most people realize. Many states have experienced higher home price appreciation than others. At the same time, states also have differences in prepayment behaviors independent of the home price index. AFT has measured these separate effects, and calculated the "intrinsic" geography factors, independent from HPI and from other loan attributes, such as loan size, that also vary between states.
"If you compare prepay speeds on otherwise identical loans from regions that have had similar HPI, for instance, Wisconsin and Texas, expected prepayment speeds will differ due to intrinsic geographic factors," Mr. Bykhovsky said.
In fact, AFT's analysis has shown that Wisconsin borrowers have the highest intrinsic propensity to refinance, whereas Texans have the lowest intrinsic propensity. In some states, such as New York and Florida, taxes on refinancing transactions also have the effect of slowing prepayment propensity.
Intriguingly, with the midterm elections just behind us, the data shows that in general, so-called blue states generally have higher intrinsic prepayment propensities than red states. AFT incorporates that intrinsic geographic factor as well as other loan and borrower level characteristics into its scoring model to project the relative prepayment propensity of loans (relative to loans of the same type, age, and weighted average coupon). That analysis can calculate the impact on the propensity to prepay on the refinancing as well on housing turnover behaviors.
"Historically there has been a vast geographic prepayment differential between different states. People assume Californians have a high propensity to refinance because prepayment speeds tend to be high there. But when you control for loan size, HPI, and all other factors, the intrinsic propensity to refinance in California is actually below average," Mr. Bykhovsky said.
He said that seven or eight factors are key to a borrower's propensity to refinance, and it is too complicated an analysis for servicers or MBS investors to "eyeball" these factors. And as home price appreciation slows, it will have different degrees of impact on different geographic areas. Housing slowdowns occur unevenly, and mortgage market participants that don't understand this, Mr. Bykhovsky predicts.
Mr. Bykhovsky says he is always perplexed when, in a changing market, he hears bankers attribute losses to "unexpectedly high prepayments."
"Unexpected by whom?" he asks. "The changes in behavior are quite predictable. They are fairly consistent." (c) 2006 Mortgage Servicing News and SourceMedia, Inc. All Rights Reserved. http://www.mortgageservicingnews.com http://www.sourcemedia.com