The Impact of HPI on Prepayments of High Credit Collateral

Introduction

by Michael Bykhovsky
President and Chief Executive Officer, Applied Financial Technology– Over the past ten years, the home price appreciation index (HPI) has undergone almost unprecedented increases – far exceeding the general inflation rate. Much has been said about the effects of an unusually high HPI on prepayment behavior. In this article we’ll examine some of the commonly held “wisdom” regarding HPI to see how it stands up to actual observations.

First, consider HPI’s effects on default rates. In spite of extremely aggressive lending practices, we have seen some of the lowest rates of default and losses for mortgage loans – an effect that is both clearly observable and understandable. However, the story changes for loans of reasonable credit quality (better than subprime, for example) that are made on prime collateral (less than 85 LTV) that constitutes most of the market. We have found the effect of HPI on these loans to be much less clear than is popularly believed. It is our view, supported by overwhelming evidence from a variety of data sources, that most of what has been said regarding the effects of high HPI on prepayment behavior is so inaccurate as to approach fantasy.

It was not so long ago that the Internet was held out as the next great revolutionary force in the world economy. We were going to transact all our business electronically, our kitchen appliances would talk to online grocery agents while office parks – and desks and chairs in general – would gradually disappear from the landscape. In the mortgage industry this meant that (finally) the refinancing process would be tamed, rendered online with perfect efficiency and economy. This would lead to total pay-downs for collateral with
even the slightest premium.

Many of our competitors accepted this “new reality” and modified their prepayment models to reflect it, only to be proven wrong shortly thereafter. Our observation is that very little has changed in refinancing response efficiency since 1993 for most of the prime collateral, once adjusted for increases in the real loan balances. Reports of ostensibly dramatic changes in the environment are, too often, used as excuses for a model’s failings.

One could argue that the increasing equity of a house allows the borrower to either obtain better loan terms or take out equity. Generally, the older the loan, the greater the equity accumulation and therefore, the greater refinancing responsiveness one would expect. But this has simply NOT occurred for the good credit/prime collateral loans we are discussing. Thus, it appears that equity accumulation has NO effect on refinancing responsiveness, and hence, HPI does not affect prepayments for prime collateral.

Other analysis examined regional differences in refinancing responsiveness. Which state would you predict as having the fastest response to refi? California? Since it is the state that has experienced by far the highest HPI that would likely be your guess. You couldn’t be more wrong. The fastest refinancing response, while controlling for all other factors (loan size, LTV, FICO, purpose residency type, etc.) is found in Wisconsin with the slowest response observed in Texas. California,

it turns out, has a housing turnover rate that is actually below the national average (when adjusted for all other factors) and refinancing rates somewhere in the 30th percentile nationally – both far below states like Wisconsin or Illinois that experienced relatively low HPI levels. So what is fact and what is fiction regarding unusually high HPI and prepayment effects? It is clear that high HPI has had a profound effect on default and loss rates. Further, a noticeable effect on refinancing response for low credit quality or high LTV collateral has been observed. However, for the remainder of the loans currently outstanding – a huge population that makes up the majority of the market – the effect of high HPI on the refinancing or housing turnover responses has not been observable.