While prepayments are expected to drop dramatically in the months ahead - given the decline in refinancings - there are still opportunities for investors in discounts as a result of strong turnover.
In a recent report from Bear Stearns, analysts recommended investors pay attention to the geographic factors when purchasing discounts. "There is a strong positive correlation between recent home price growth and housing turnover activity, with every 10% of home price appreciation adding almost a full CPR to the housing turnover level," they concluded. After reviewing prepayment data from 2000, as well as home price growth over the past five years, Bear Stearns predicts that housing turnover prepayments in the 2004 to 2005 period will exceed 2000's experience by about 15%.
The regions of the country that are predicted to record the highest housing turnover levels are California and Nevada, followed by the New England states and Maryland. Florida and Virginia are also expected to be in the top one-third of the prepayment distribution due to very strong home price growth in the past five years.
Unlike 2000, Colorado and Arizona are expected to sink towards the middle of the pack, says Bear Stearns. Most of the mountain and southern states, the Pacific coast region, Texas and New York are expected to drift to the bottom third in state prepayment rankings.
JPMorgan Securities has also been suggesting that discount speeds will be faster than expected due to higher turnover. Some reasons suggested are the increased move to ARMs as rates have risen; the historically low mortgage rates; and continued strong home price appreciation.
ARM popularity is not enough
Lehman, however, is "not persuaded that the growing popularity of hybrid ARMs is enough." Aggregate borrower mobility, analysts said, will be different than in previous rising-rate periods. "On the flipside, there are some very convincing reasons why turnover (post-summer) should settle at levels lower than that seen during 2000," analysts wrote.
The firm's first point deals with the use of the MBA Purchase Index as an indicator of turnover. Lehman does not believe it is a good indicator because there is measurement bias in the Purchase Index. Lehman is not convinced that the MBA accounts for the consolidation in the banking industry accurately. Furthermore, the Purchase Index takes into account new- and existing-home sales. However, only existing- home sales contribute to turnover, says Lehman. The growth in this index since summer 2000 has been about half that of the Purchase Index.
Lehman also notes that as the economy continues to strengthen, higher mortgage rates are likely to dampen overall housing activity. So once the summer is over, analysts expect prepayments on discounts to gradually converge to historical averages. This has few implications for mortgage valuations. "To the extent that turnover comes in 25% higher than our forecasts but reverts to model predictions over the coming six months, valuations of typical mortgage cash flows do not change significantly," Lehman said.
Lehman analysts also question whether the purchase market can sustain current levels. The question is not where discounts prepay over the summer, but how they will prepay over the long term. Researchers are not convinced that borrower mobility in the future will be substantially greater than in the past. Also, they don't buy the argument that the growing popularity of ARMs, including hybrids, has created a structural shift in the market leading to higher turnover than in the past. For starters, states Lehman, ARM share has been at similar levels as prior purchase-market environments. In the early 90s, and in the 1999-2000 period, the proportion of new mortgages in ARMs was more than 40%. "So why should a higher share of ARMs today completely change the dynamics of the housing market?" asked analysts.
Therefore, Lehman concluded that given the current steepness of the curve, as well as the length of time it has been steep, "borrower self-selection should have a significant dampening effect on turnover for 2003 and 2004 origination 30-year collateral." They say the slope at origination for 2003 and 2004 4.5s through 5.5s implies a 1.5% to 2% CPR lower turnover rate than that experienced in the 1990s. However, accounting for the high concentration of refinancers in recent cohorts, this would add at most 0.5% CPR to unseasoned cohorts. A more realistic assumption, says Lehman, is for turnover to drop off more than what was experienced in the late 1990s as the Fed starts raising rates. Analysts currently estimate 2% to 3% CPR lower current coupon prepayments and 1% to 2% CPR lower prepayments on deep discounts relative to the late 90s.
Copyright 2004 Thomson Media Inc. All Rights Reserved.