Excerpted from an eight-page report issued December 14, by Brian Lancaster, managing director and head of CMBS research at Bear, Stearns & Co.
"However, investment-grade CMBS tend to widen more than non-callable corporate bonds as their price increases over par. The reason is investor concern over a default-induced prepayment."
That quote, from our report on how CMBS traded in early 1999, is still unbelievably enough, true today. Whether it is because a number of CMBS participants come from the residential MBS market where price compression and spread concessions for higher priced bonds are a way of life, or whether it is a holdover from the days when prepayment protection in the CMBS market was weak, this "rule of thumb" makes no sense for many CMBS today. Indeed, we believe premium, second-tranche CMBS represent the best relative value in the AAA CMBS market.
Regardless of their position in a deal, AAA CMBS priced over par commonly trade at a spread several basis points wider than if they were priced at par. The general "rule of thumb" currently is that a high quality 10-year AAA CMBS would trade 3 bps wider if it had a 103 handle than if it had a par handle. The spread would widen by about another 2 bps/point until a 105 handle after which the spread would widen by about 3 bps per point (see table below). In the five year sector, the "rule of thumb" is 3 bps wider for a 103 handle and about another 3 bps/point until a 105 handle after which the spread would widen by a similar amount.
As noted above, the principal reason for this in deals originated in the last couple of years is the possibility of a default-induced prepayment. While the possibility exists of a voluntary prepayment, the likelihood of the investor receiving a voluntary prepayment or one without offsetting compensation on recently originated deals is low. The quality of prepayment protection on deals originated in the last two years has improved dramatically. For example, 79% and 85% of the loans in 1999 and 2000 vintage CMBS deals were protected from prepayment by lockout/defeasance. Another 20% and 13%, respectively, were covered by yield maintenance. This is in sharp contrast to 1996 when only 28% of the loans in CMBS issued that year were protected by lockout/defeasance and 58% were protected by yield maintenance.
To underscore the quality of prepayment protection on recent deals and to show that there is little difference in the way prepayments impact the yield and average life of recent vintage deals priced either at a discount or premium, we did the following. We stressed bonds from three varied deals issued this year, priced at premiums ranging from almost 4 points to over 6 points as well as a number of bonds trading at 2 to 3 point discounts. We first assumed no prepayments (0 CPR in the base case), then 100% CPR after lockout but during the yield maintenance period, if there was one.
The yields and average lives of the second-tranche bonds, whether a discount or a premium, hold up quite well under these extreme prepayment scenarios.
Certainly, from a prepayment perspective, it would seem that the miniscule difference in yield and average life performance of these bonds, whose only difference is that they are priced at a high premium and discount, does not justify the premium bonds being priced 10-20 bps cheaper.
Impact of Default-
So if the risk of voluntary prepayment adversely affecting premium bonds is low then what is the likelihood of a default-induced prepayment impacting a 5-year or 10-year AAA CMBS? In the case of the former it is significant. In the case of the 10-year AAA sector, the risk is negligible. The reason for this is that second tranches in CMBS deals typically receive no principal payments, default induced or otherwise for a number of years. In contrast, first tranche AAA CMBS typically begin receiving principal immediately.
To show this, we stressed the same premium and discount tranches with increasing levels of defaults. We assumed that in one year, 2% of the outstanding loans on BSCMS 2000-WF1 default with no lag to recovery and 30% severity. There was no impact on the yield of the 10-year AAA A2 tranche, which was priced at 106:30 and its average life barely budged from 8.78 years to 8.77 years. We then doubled, tripled, quadrupled and quintupled the amount of defaults to 10% with no impact on yield and a tiny impact on average life! Finally, even at the extraordinary level of 20% defaulting ,the yield edged down by only 2.5 bps and the average life shortened by only .24 years.
The same was true within a few basis points for the other second-position AAA tranches we stressed, as well as the second position discount AAA tranches. Clearly, there was little if any benefit here to being priced at a discount versus a premium. To put into perspective the extreme nature of these scenarios, consider a widely regarded study that found cumulative defaults for commercial real estate over a 10 year period covering severe market conditions, averaged 18% with severities from 19% to 38% .
Extension Risk May be More Likely than Call Risk
Given the ability of premium second-tranche AAA CMBS to withstand these kind of stresses, we believe that the drastic spread widening as they trade over par is unjustified. Indeed, extension risk may be more pertinent to these tranches than call or default risk. Thus, in theory, premium-priced CMBS could be more attractive than par or discount bonds.
Large amounts of commercial mortgage loans must be refinanced in the next 6 to 7 years. If rates rise to levels higher than the refinancing constants incorporated in bond ratings by the rating agencies, then not only could defaults rise which would affect the first tranches through default-induced prepayments, but borrowers most likely would need additional time to refinance their loans. The latter could cause CMBS to extend which, given that they are trading at a premium, would help offset their average life extension by providing the investor with a small yield pickup. Discount CMBS, of course, would suffer doubly under this scenario as their average lives increase and yields decline.
Component Pricing Shows Premiums to be Underpriced
Finally, to confirm our belief that second-position, premium-priced CMBS are cheap versus par CMBS we did the following. We compared the price of the A2 tranche of BSCMS 2000-WF1 (105:03 as of 12/4) intact versus the combined price of the bond if it were split into a par or near-par priced bond and an IO off the same cash flows. BSCMS 2000-WF1 A2s would be priced at approximately swaps plus 36 bps, if they were a par or near-par priced security (see table below). The IO off these cash flows would be priced at approximately 75 bps over the 8.79-year interpolated swap at 100 CPR for a yield of 7.39% and a price of 4.8550 . Given the price of the two components individually, the implied price of the two together is 105.3328 or 105:11, which is equivalent to swaps plus 38 bps. In contrast, the price of BSCMS 2000-WF1 A2 in the market was actually 105:03 (swaps plus 42 bps) which is about 8/32nds or 4 bps too cheap!
Premium-priced AAA CMBS bonds currently trade at wider spreads than if they were trading at par. While this may be justified in the case of first-tranche bonds, we believe it is unjustified in the case of second-tranche AAA CMBS bonds, particularly newer ones. Extreme stress tests with voluntary prepayments and default-induced ones show little dif-ference in the way they impact premium and discount second-tranche CMBS. Moreover, component pricing, whereby the premium is split into a par-priced bond and an IO off the same cash flows, shows premiums to be 8/32nds or 4 bps cheap vs. par bonds. In addition, extension risk due to potential refinancing issues looms in the future. We believe that discount bonds trading substantially tighter than par bonds are rich, and premium-priced bonds trading substantially wider than par bonds are cheap.