With the proliferation of non-traditional products in the mortgage market, analysts are starting to look at the credit implications of such growth.
In a recent report, UBS analyzed the layered risk of these creative mortgages, including hybrid ARMs that have a variety of innovative features, such as interest-only, negative amortization, and silent second loans.
In its study, UBS chose to focus on IO loans since they have the longest history. Also, as part of the increase in "creative mortgages" or "affordability products" in the non-agency market, the IO hybrid sector has seen significant growth over the past year as borrowers try to more effectively manage their monthly mortgage payments. This has understandably raised questions about their expected credit performance especially when these loans reset or begin to amortize.
To illustrate the growth of IOs, UBS said that for the Jumbo and Alt-A MBS sectors, the market share of IO mortgages have tripled to 60% to 70% from approximately 20% at the end of 2003. Additionally, in subprime mortgages, the IO share has quadrupled, to currently being about 20% from just 5%, at the end of 2003, with no signs of slowing down.
In studying the credit performance of these loans, analysts noted that the short performance history and robust housing sector of late has made for very low loss rates for creative mortgages, specifically IO collateral. Analysts said that for states like California, for instance, many prime transactions do not have any reported losses.
Data illustrated that IO loan performance improves as one moves down in credit, reflecting the trend noted by researchers of creative mortgages expanding in influence from the high credit to lower credit borrowers, or from the prime to supbrime sectors.
Prime lenders initially test these products on their most prime customers and then go down to their average prime borrower and high-end Alt-A clients. As soon as these lenders feel more comfortable about a new product offering, they go further down in credit and offer similar mortgages to their subprime clientele.
Analysts report that it is hard to impose very strict underwriting standards for prime quality borrowers. They added that even if FICO scores on these types of creative mortgages were higher, the performance impact would not be as considerable as it would be in the subprime world.
"Prime credit borrowers tend to have a clean credit history and strong financial capacity, so it is difficult to ask for compensating factors," explained analysts. "As the lenders move down in credit, there is more room to maneuver, because lenders simply reserve the new creative mortgage products for the top borrowers among their applicant pool."
UBS said that the widening performance gap observed at the low end of the credit spectrum shows that there is a strong selection bias in the first life stages in creative mortgages. In short, analysts stated that performance histories of subprime creative mortgages are "distorted by strong selection bias." However, this bias diminishes as these types of mortgages become more popular. Minus this selection bias, the analysts said the higher leverage would probably manifest itself as a negative factor in collateral performance. Although the ultimate test, they said, is what would happen when these creative mortgages reach their reset date, as a majority of these are hybrid ARMs, or when the housing sector slows.
In a separate report, Countrywide Securities also looks into the credit performance of IO loans. To study IO loan credit performance, Countrywide Securities used the limited available data to analyze the sector. Countrywide first looked at borrower attributes of recently-issued Countrywide hybrid ARM loans. Analysts found that in terms of conforming hybrid loans in aggregate, average FICO scores were slightly higher on IOs versus non-IOs, and the same holds for average LTVs and average combined LTVs (CLTVs) as well as for the average loan amount. In addition, IOs tended to be purchase loans while more non-IOs were refinancings. The results were similar for Jumbo loans.
Countrywide believes the last two issues - higher purchase percentage and higher combined LTVs - are particularly important in understanding the performance of IO loans and interpreting the data that is currently available. One piece of data they found useful is cumulative 60-day delinquencies, which is considered a leading indicator of defaults and losses. Analysts noted that IO hybrid ARMs have experienced marginally higher delinquencies than non-IO loans.
Looking further into the numbers, Countrywide noted that the IO loans have more second liens. The use of second liens issued at the time of origination is a popular option for borrowers preferring not to make a 20% down payment. Taking out a second mortgage eliminates the need for paying mortgage insurance. The results show that the presence of a second lien impacts the delinquency performance of IO loans more than non-IO loans. Countrywide reports that this is noteworthy since IO loans have a greater concentration of purchase loans. There are two explanations for this behavior. First is the presence of a second lien indicates a more financially stretched borrower. Alternatively, there may be a self-selection-effect taking place, where weaker borrowers are more inclined to take out a "simultaneous second." Countywide estimated that purchase loans with accompanying second liens comprise around 20% of the IO ARMs originated.
Another important issue is the post-reset behavior. The sector does not have meaningful historical performance data, so Countrywide reviewed post-reset behavior on 3/1 hybrids to draw some inferences. In a historical review of cumulative defaults and 60-day delinquencies, analysts reported both that the numbers are low; and that both measures do not begin to level off until well after the reset (month 36). In addition, Countrywide analysts noted the cumulative defaults in their study do not appear to level off much at all. The data suggests that higher credit borrowers ultimately refinance out of the pool. Still, as far as the 3/1 history shows, post-reset increases in credit problems seem to be well contained.
Countrywide also evaluated losses for hybrid IO ARMs assuming different prepayment speeds and post-reset loss rates. The analysis demonstrates the significant impact of prepayments on the relative performance of IO ARMs, and suggests that in the context of prepayment expectations incremental losses on IO ARMs (relative to non-IO ARMs) can be expected to remain relatively low.
Copyright 2005 Thomson Media Inc. All Rights Reserved.