With the surge in hybrid ARM issuance, mortgage-backed analysts are starting to change their models to reflect this phenomenon - shifting their emphasis from a reliance on historic data to being more borrower-oriented. This change is in line with the methodology they are currently using in their agency and non-agency fixed-rate models. Notably, hybrid non-agency and agency combined issuance is estimated to reach $400 billion in 2003, nearly double the pace of last year.
In a recent Lehman Brothers report, analysts also urge MBS investors to look at non-index sectors, including hybrid ARMs, to outperform the benchmark index.
A Bear Stearns report on the firm's new hybrid mortgage prepayment model said that the established practice in prepayment modeling was to treat the borrower as merely a generic entity, and to align forecasts to historical data.
"We argued that this approach had become too simplistic given the granularity of detail used by investors to distinguish value in today's multi-sector mortgage market," the Bear report said. "Hence, we revamped our methodology, shifting the emphasis from data fitting to a more thorough modeling of borrower behavior."
This is accomplished by identifying and including variables that might directly or indirectly quantify borrower characteristics; such characteristics determine whether these borrowers would refinance. For instance, borrowers with higher loan balances have more of a dollar incentive to refinance, while those with more accumulated home equity have a greater propensity for a cash-out refinancing.
The use of borrower attribute-based methodology on Bear's agency and non-agency fixed-rate models was the basis for the hybrid model that was presented in the report.
Hybrid borrowers can be put into three main groups, according to the report. The first are "traditional" hybrid borrowers, who select this type of loan because they have an expected tenure in the home - three, five or seven years - or a planned refinancing horizon. The second group is made up of "rate refinancers," who choose a hybrid mainly for the rate-risk compensation provided by the product's lower coupon. The third group is composed of "first-time homeowners," who utilize the relatively low monthly payment provided by hybrids as a way to qualify for a mortgage. Though it was a small category this year, Bear expects to see more borrowers using the hybrid as a way to qualify for a mortgage in 2004. This is an effect of the market's shift from a refinancing to a purchase origination environment.
Projected monthly prepayments in Bear's new model would result from the sum of two sub-models - a refinancing sub-model and a housing turnover sub-model. The refinancing model covers the effects of the following: WAC-to-current rate differentials, loan size, equity accumulation, threshold-media events generated by historically low rate levels, rate premiums at origination, burnout, and the shape of the yield-curve.
Meanwhile, the housing turnover sub-model forecasts prepayments that are caused by: demographics, loan seasoning, seasonal cycles, and home price appreciation.
Housing turnover sub-model
Dale Westhoff, senior managing director at Bear Stearns, said the housing turnover component of the new hybrid model is the most important factor that distinguishes it from traditional models.
"Hybrid borrowers are a self-selected group," said Westhoff. "Many of them know that they are going to be out of their houses within the specified three-, five-, or seven-year period." Hybrid borrowers have a greater propensity to change houses within the fixed rate term of the hybrid, leading to higher housing turnover levels compared to those found in the 30-year sector (where borrowers are expected to stay in their homes for a much longer period).
The seasoning profile is driven by the hybrid's fixed-rate timeframe, whether it is three, five, or seven years. "In general, the shorter the time to first reset, the faster the seasoning process and the higher the fully seasoned housing-turnover prepayment," Westhoff explained in the report.
Westhoff added that hybrids ramp up much faster than their 30-year fixed-rate counterparts; which is an extremely important factor in the next year given higher interest rates and growing extension fears.
Additionally, the comparatively fast prepay rates in hybrids (resulting from the relatively fast rates of housing-turnover seasoning on hybrid pools) is the source of the extension risk protection offered by hybrid mortgage-backeds. The report said that as rates back up, hybrid prepay speeds drop less compared to 30-year fixed-rate loans. This is why the extension of hybrid average lives is relatively low versus fixed-rate product.
"We expect hybrid MBS to appeal to a lot of mortgage investors in 2004 because the sector offers significant protection from extension risk," said Westhoff. "Liquidity in the sector should also be supported by an increase in hybrid ARM market share as we transition from a refinance to a purchase origination environment."
On the refinancing part of the model, Bear focused on modeling the multiple rate incentives available to hybrid borrowers - particularly looking at borrower incentives to move into shorter- or a longer-term hybrid ARM product or whether to move into a 30-year fixed-rate mortgage.
According to the report, the principal objective of the model is to maximize the accuracy in valuating and assessing the risk in hybrid ARMs. "In this effort, the overriding concern is whether borrowers refinance; which products give them an incentive to refinance is a secondary concern," analysts wrote.
For instance, to predict prepayments on a pool of 5/1 ARMs, the model only needs to incorporate the relevant rate incentive, it does not need to consider how many of these borrowers are actually refinancing into another 5/1 hybrid compared to those refinancing into a 3/1 hybrid or versus a 30-year fixed-rate mortgage.
The Lehman report
Meanwhile, Lehman Brothers also presented a new hybrid ARM prepayment model. With the recent sell-off and the increased homogeneity in mortgages, investors should look at non-index sectors such as hybrid ARMs to outperform the benchmark index. "From this perspective, we felt the need for a robust pricing tool that will enable investors to identify relative value opportunities," wrote analysts.
The Lehman model also focuses on the two aspects of refinancings and turnover. On the refinancing aspect, Lehman's prepayment model is based on the refinance experience of January 2002 to June 2003. The model shows that agency hybrid prepayments are better behaved compared to fixed rates. Meanwhile, on the turnover front, hybrids have reflected much faster prepay speeds compared to fixed rates. This is due to the high percentage of hybrid borrowers having a short horizon. For turnover, Lehman used the 1994 to 1995 discount period.
Additionally, in Lehman's model, hybrids seem compelling compared to their fixed-rate counterparts at a pick up of 25 to 30 basis points in OAS.