(Bank of America) Hybrid ARM MBS - Valuation and Risk Measures
(Bank of America) Hybrid ARM MBS - Valuation and Risk Measures
(Bank of America) Hybrid ARM MBS - Valuation and Risk Measures
Hybrid ARM MBS: Valuation and Risk Measures November 27, 2006
Sharad Chaudhary Hybrid ARMs have emerged as a popular investment alternative for investors who
212.583.8199 seek better extension protection and/or wider spreads than those offered by fixed-
[email protected]
rate mortgages. Institutional investors participating in the Hybrid ARM MBS sector
include banks, REITS, insurance companies, money managers and hedge funds.
RMBS Trading Desk Strategy Prices of Agency hybrid ARMs are generally quoted in terms of Z-spread, which is
Ohmsatya Ravi a cash flow spread to the implied spot Treasury curve. The two exceptions to this
212.933.2006 quoting convention are 10/1 and seasoned hybrids ARMs. We present a detailed
[email protected]
discussion on the different quotation systems used in the ARMs market.
Qumber Hassan The subject of tail valuation in hybrids has received significant attention recently
212.933.3308 due to an unprecedented level of discount seasoned hybrids resetting in 2006 and
[email protected]
2007. We present a detailed discussion of several factors that impact tail valuations.
Sunil Yadav The Z-spread, OAS to reset and Bond Equivalent Effective Margin (BEEM)
212.847.6817
[email protected]
measures have some significant limitations for the relative value analysis of ARMs.
OAS to maturity is a much better relative value metric since it accounts for the fact
Ankur Mehta that the tail is not worth par, cap structures have a non-zero value, and that
212.933.2950 prepayment speeds are not independent of rates.
[email protected]
We provide a detailed discussion on the risk exposures of hybrid and fixed-rate
MBS. Hybrid ARM MBS provide a better hedge against extension risk relative to
RMBS Trading Desk Modeling fixed-rate products. In addition, short reset hybrids are a good hedge against
ChunNip Lee housing slowdown.
212.583.8040
[email protected]
Unlike fixed rate pass-throughs that have a fixed coupon and a variable servicing
fee, the vast majority of hybrid ARMs pay a weighted average coupon (WAC) to
Marat Rvachev security holders and have a fixed servicing fee. The presence of a fixed servicing
212.847.6632 fee in hybrid pools has the effect of allowing their WACs to change over time
[email protected]
depending upon how the individual loans within a pool pay down. An in-depth
Vipul Jain analysis of the factors that influence WAC drift and the implications from the
212.933.3309 valuation standpoint are presented in this primer.
[email protected]
This document is NOT a research report under U.S. law and is NOT a product of a fixed income research department. This document
has been prepared for Qualified Institutional Buyers, sophisticated investors and market professionals only.
To our U.K. clients: this communication has been produced by and for the primary benefit of a trading desk. As such, we do not hold
out this piece of investment research (as defined by U.K. law) as being impartial in relation to the activities of this trading desk.
Please see the important conflict disclosures that appear at the end of this report for information concerning the role of trading desk
strategists.
RMBS Trading Desk Strategy
I. INTRODUCTION
Hybrid ARMs have emerged as a popular investment alternative for investors who seek
better extension protection and/or wider spreads than those offered by fixed-rate mortgages.
A wide range of choices in selecting the length of the initial reset period in hybrids allows
investors to tailor their investments according to their duration preferences. Furthermore, as
discussed in our earlier publication on hybrid ARMs,1 because of the shorter tenure
horizons of the underlying borrowers in hybrid ARM pools, hybrid ARM MBSs possess
better convexity characteristics than fixed-rate MBSs. Consequently, hybrid ARMs not only
provide excellent extension protection when rates backup, they also provide better
prepayment protection than fixed-rate MBSs in a heavy refinancing environment. Recently,
Lehman Brothers announced their intention to include more than $300 billion Agency
Hybrid ARMs in their U.S. Aggregate index starting April of next year. This is expected to
deepen the participation of institutional investors in the ARMs market even further.
In view of the growing popularity of hybrid ARMs, this primer details some of the nuances
in Hybrid ARM MBS valuations, analyzes the risk profiles of ARMs, and provides some
perspective on relative value issues. First, we define and discuss the standard quoting
convention for hybrids and address some of the issues with using nominal spreads to gauge
relative value in hybrids. This section also includes a brief discussion on why OAS is a
superior relative value tool and presents an example to demonstrate how nominal spreads
can give quite different (and often wrong) relative value signals than OAS.
In the next section, we review the fundamentals of tail valuation in hybrids. To simplify the
discussion on tail valuation, we break down the problem into two parts by first considering
tail valuation of hypothetical securities without any caps and then focusing on the valuation
of embedded caps.
The discussion then segues into an analysis of the risk exposures of hybrid ARMs in
relation to fixed rate mortgage products. We highlight the key differences between hybrid
ARMs and 15-year and 30-year fixed rate MBSs in terms of duration, convexity, curve and
volatility exposures.
Next, we present an overview of hybrid ARM investors and their motivations for investing
in the hybrid ARM sector. Wider spreads make hybrid ARMs appealing to a wide range of
market participants. In addition, investors who are concerned about extension risk or a
housing slowdown also prefer hybrids over fixed-rate MBS.
Our primer concludes with an Appendix, presenting a detailed discussion on the issue of
WAC drift in hybrid ARMs. The presence of a fixed servicing fee in hybrid pools has the
effect of allowing their WACs to change over time depending upon how the individual
loans within a pool pay down. An in-depth analysis of the factors that influence WAC drift
and its implications from the valuation standpoint is presented in this primer.
1
Please see the primer titled Prepayments on Agency Hybrid ARM MBS published on November 13, 2006.
2
RMBS Trading Desk Strategy
Z-Spreads
The standard quoted Z-spread is calculated assuming that the underlying collateral will
prepay at 15% CPR during the initial fixed-rate period of the security and that the borrowers
will make a balloon payment of the remaining principal balance at the end of the fixed-rate
period. The pricing convention that a hybrid pays off all its remaining outstanding balance
at its first reset date is called the CPB assumption. In reality, not all hybrid borrowers
prepay the remaining principal balance at the end of the fixed-rate period. Because of this, a
hybrid security holder continues to receive a series of principal and interest cash flows after
the first reset instead of receiving all the remaining principal back at the time of the first
reset. From a valuation perspective, the CPB assumption implies that the value of the series
of cash flows received after the first reset (called the “tail”) is equal to the face value of the
security at the time of the first reset (i.e., the tail is worth par).
Although the Z-spread is a commonly quoted number, investors need to be aware that it is
not a good relative value tool because of the following reasons:
• The Z-spread is calculated assuming that the tail is worth par. In practice, not all
Hybrid ARM borrowers prepay their mortgages before or at the reset date.
• The Z-spread does not capture the value of different cap structures. The Z-spread
values a 2/2/5 cap the same as a 5/2/5 cap or a 2/2/6 cap.
• Z-spreads are not good relative value indicators because they change with the level
of interest rates and volatility. Note that Z-spread calculations use a 15% CPB
assumption while actual prepay speeds change with the level of interest rates.
2
Recently, dealers have started quoting Agency 10/1 hybrids on a Z-spread basis to allow for more consistent pricing.
3
Months-to-Roll.
3
RMBS Trading Desk Strategy
Similarly, prepayment and cap risks involved in a hybrid ARM pool change when
implied volatilities move. Consequently, the same Z-spread indicates different
levels of risk premium at different rate and volatility levels.
4
Bond-equivalent yield on a pass-through security is computed so that it is comparable to a yield computed on a coupon security paying semi-annual
interest, calculated to the settlement date.
4
RMBS Trading Desk Strategy
Figure 1: Z-Spreads of Benchmark 5/1 Hybrid ARM Security vs. Level of Rates
85 5.4
80 5.2
5
75
Z-Spread (bps)
Yield (%)
4.8
70
4.6
65
4.4
60 4.2
55 4
11/02/05
12/02/05
01/03/06
02/01/06
03/02/06
03/30/06
04/28/06
05/26/06
06/26/06
07/25/06
08/22/06
09/20/06
FNAR 5.50% 5/1 10-yr Tsy Yield
Figure 2: OAS of Benchmark 5/1 Hybrid ARM Security vs. Level of Rates
25 5.4
5.2
20
5
15
OAS (bps)
Yield (%)
4.8
4.6
10
4.4
5
4.2
0 4
11/02/05
12/02/05
01/03/06
02/01/06
03/02/06
03/30/06
04/28/06
05/26/06
06/26/06
07/25/06
08/22/06
09/20/06
5
RMBS Trading Desk Strategy
From the figure we see that seasoned hybrids trade through new hybrids in terms of Z-
spread. This however, does not reflect any relative value opportunity. The tighter spread on
the seasoned hybrid merely reflects the fact that a higher percentage of its value is coming
from its tail, which the Z-spread fails to capture due to its construction. The Z-spread is
based on 15% CPR to balloon date and hence, completely overlooks the tail contribution.
Figure 3 also shows the BEEM for the selected bonds based on a 15% CPR for life as well
as prepayment projections from our model. Notice how significantly the BEEM declines as
we use a more realistic prepayment vector in place of 15% CPR for life. From a relative
value perspective, the inability of BEEM to account for the option cost results in securities
with lower option cost appearing rich relative to those with higher option cost (all else being
equal). For example, the 5.5% seasoned 5/1 hybrid ARM (36 MTR) with a 5/2/5 cap
structure looks a few bps tighter than the corresponding bond with a 2/2/5 cap structure in
terms of BEEM.
Finally, OAS to reset has similar limitations as the Z-spread since it ignores the contribution
of the tail. The seasoned 5/1 hybrid ARMs (36 MTR) with a 5/2/5 cap have a tighter OAS
to reset than the corresponding 5/1s with a 2/2/5 cap across the three coupons, even though
all of them have same OAS to maturity. This occurs as a result of the higher tail valuation
for the former group due to a favorable cap structure. In this case, the tighter OAS to reset
on the 5/1 ARMs with a 5/2/5 cap structure does not imply richer valuations; it simply
reflects the limitation of OAS to reset in reflecting the higher tail value on the 5/2/5 cap.
Similarly, the 5.5% seasoned 5/1 hybrid ARM (36 MTR) has a tighter OAS to reset than the
newly issued 3/1 hybrid ARM in spite of having a wider OAS to maturity. Again, the tighter
OAS to reset compensates for the higher tail valuations on the 5/1s and is not an indicator of
richer valuations. A more fundamental question while analyzing the tail valuations of these
securities is the reason behind the higher tail value on the 5/1s. Even though both securities
have 36 months-to-roll, they have a different prepayment profile. Seasoned 5/1s will
initially prepay faster than the new 3/1s; however, as the 3/1s season, their prepayment
speeds will be faster than the 5/1s resulting in a smaller factor at reset. The combination of
higher baseline speeds and the smaller factor at reset on 3/1s reduces its tail value when
compared to seasoned 5/1s. This difference in the tail value between the 3/1s and the
seasoned 5/1s fades from 4 ticks on the 5.5% coupon to 2 ticks on the 4.5% coupon. The
reduction in tail value difference as we move down from the 5.5% coupon to the 4.5%
coupon occurs because the likelihood of the 3/1 and seasoned 5/1 ARMs hitting their cap
increases (since they both have a 2% initial cap). Both these securities may become a
discount if their rates hit the cap and because the factor on the 5/1 ARM is more than that
on the 3/1, it hurts the 5/1 tail valuations more. The likelihood of hitting the cap reduces if
the cap structure changes to a 5/2/5 cap. This is why the 5/1 ARM with the 5/2/5 cap
structure has a fairly stable tail value across different coupons.
To summarize, a tighter OAS to reset does reflect a higher tail valuation but not necessarily
a richer tail valuation. OAS to maturity is a more comprehensive measure of relative value
and we recommend using it over Z-spread, BEEM and OAS to reset - all of which fail to
fully capture the tail value in a hybrid ARM.
6
RMBS Trading Desk Strategy
7
RMBS Trading Desk Strategy
• First, the outstanding principal balance at the first reset and how it reduces over the
remaining life of the underlying mortgage through scheduled and unscheduled
principal payments will naturally influence the total cash flows due to this IO strip. All
else being equal, the higher the outstanding balance at the first reset and the slower
subsequent prepayment speeds, the higher will be the value of the IO strip.
• Second, the size of the net margin is obviously important. All else equal, the higher the
net margin, the higher the total cash flows due to the IO strip. However, the tail value
generally increases by less than what one would expect for a given increase in the net
margin because a higher margin usually leads to faster prepayments on the collateral.
• Third, although we ignored caps and floors in this simplified example, they will impact
the value of the IO strip in actual hybrid securities because caps and floors can
effectively alter the cash flows due to the IO strip.
So far, we have considered the value of the IO strip at first reset. i.e., how far above par is
8
RMBS Trading Desk Strategy
the tail worth at the time of first reset? Now, let us consider the value of all the tail cash
flows at the time of buying a hybrid security. The tail cash flows on the hybrid security may
be thought of as consisting of two components: a security paying back the remaining
outstanding principal on the first reset date and a forward IO strip which starts generating
cash flows after the first reset. Usually, market participants refer to the value of the IO strip
discounted to the time of the purchase of a security as the “tail value”.
The tail value, as defined above, usually decreases as the length of the first reset period
increases. This is largely because of the lower remaining balances of longer hybrids at first
reset. The discounting of tail cash flows over a longer period of time also reduces the value
of the tail on longer hybrids relative to the value of the tail on comparable short hybrid
products. Thus, 3/1s usually have higher tail values than 5/1s and 7/1s, and 5/1s have higher
tail values than 7/1s.
Figure 4 shows our estimates of the value of the tail of several hybrid securities. Note that
the “tail value” here simply means the above par value of the tail at first reset discounted to
the time of purchase of the security. It can be estimated using either of the following two
approaches: a) the difference between the OAS computed with and without the balloon
payment assumption at first reset; or, b) the difference between the constant OAS prices of
a hybrid security with and without the balloon payment assumption at first reset. In the first
approach, the dollar price of the security is kept constant and two OAS values referred to as
“OAS to reset” and “OAS to maturity” are calculated. The “OAS to reset” is calculated
using model determined prepay vectors for the time until the first reset followed by a
balloon payment at the first reset. The “OAS to maturity” is calculated using model
projections of prepayment speeds until final maturity of the security. The tail value can then
be obtained by taking the difference between the “OAS to maturity” and the “OAS to
reset”. In the second approach, the “OAS to maturity” of a hybrid security is calculated first
at market price of the security and then the price of the security is recalculated at the same
OAS, but assuming that the remaining outstanding balance will be prepaid at first reset
(“Price to reset”). The difference between the market price and the price to reset represents
the “tail value”.
9
RMBS Trading Desk Strategy
10
RMBS Trading Desk Strategy
increases the value of embedded caps. Similarly, implied 1-year forward rates at the time of
reset also change when length of the fixed-rate leg changes, which in turn has an impact on
the value of embedded caps.
Finally, the actual cap structure itself affects the value of a cap. Clearly, a 2/2/5 cap will
have higher value than that of a 5/2/5 cap on the same collateral. In other words, a security
with a 2/2/5 cap will be worth less than a security with a 5/2/5 cap. Note also that the
difference between the values of the two cap structures keeps changing as interest rates
move. Figure 5 shows our model based pay up for different coupon 5/1 hybrids with a 5/2/5
cap structure vs. their 2/2/5 counterparts. Note that the pay up for higher initial cap
decreases as the coupon increases.
11
RMBS Trading Desk Strategy
12
RMBS Trading Desk Strategy
the short term rates decline more than the long term rates), hybrids appreciate
more in price relative to15-years.
4. Volatility Exposure: Figure 6 shows volatility durations of hybrids and fixed-rate
products. The volatility duration shown here is defined as the change in the price
of a security (in cents) per 1 bp change in the implied volatility of a representative
swaption. The volatility exposure of hybrids is strongly dependent on the length of
the initial reset period and usually increases with the length of the reset period.
The volatility exposure of hybrids also differs from that of 15-year and 30-year
fixed-rate products because they are exposed to different parts of the volatility
surface. For instance, the volatility exposure of current coupon 30-year fixed-rate
mortgages is frequently hedged using 3yr*7yr, 3yr*10yr or 5yr*10yr swaptions.
The appropriate swaptions to use for hedging a hybrid security will have shorter
expiration and maturities than the swaptions used for hedging fixed-rate
mortgages. The presence of caps adds an additional complexity to hedging the
volatility exposure of hybrids. For instance, a first reset cap on a 5/1 hybrid is
nothing but a payer swaption on a 1-year rate with an option expiration of 5 years
(i.e., 5yr*1yr swaption) as explained before.
13
RMBS Trading Desk Strategy
Figure 7: Key Rate Partial Durations of Hybrids and Fixed-rate MBS (10/12/06)
Partial Duration
Coupon Price
2 Year 5 Year 10 Year 30 Year
3/1 5.00 99.34 1.42 (74%) 0.61 (32%) -0.04 -(2%) -0.08 -(4%)
5.50 100.23 1.42 (86%) 0.44 (27%) -0.10 -(6%) -0.11 -(7%)
6.00 101.04 1.39 (101%) 0.29 (21%) -0.16 -(11%) -0.15 -(11%)
5/1 5.00 98.67 1.01 (38%) 1.73 (65%) 0.04 (2%) -0.14 -(5%)
5.50 99.93 1.09 (51%) 1.35 (63%) -0.12 -(5%) -0.20 -(9%)
6.00 100.86 1.12 (68%) 0.99 (60%) -0.24 -(15%) -0.23 -(14%)
7/1 5.00 97.86 0.84 (25%) 1.85 (55%) 0.83 (25%) -0.17 -(5%)
5.50 99.39 0.97 (36%) 1.52 (56%) 0.45 (17%) -0.23 -(9%)
6.00 100.55 1.04 (50%) 1.15 (56%) 0.14 (7%) -0.27 -(13%)
10/1 5.00 96.95 0.72 (17%) 1.34 (32%) 2.26 (54%) -0.11 -(3%)
5.50 98.89 0.88 (26%) 1.15 (34%) 1.52 (45%) -0.20 -(6%)
6.00 100.32 0.99 (39%) 0.91 (36%) 0.86 (34%) -0.25 -(10%)
15 Year 4.50 95.91 0.64 (15%) 1.53 (37%) 1.87 (46%) 0.07 (2%)
5.00 97.77 0.75 (20%) 1.41 (38%) 1.50 (41%) 0.02 (0%)
5.50 99.53 0.87 (29%) 1.21 (40%) 0.99 (33%) -0.07 -(2%)
6.00 101.13 0.97 (43%) 1.01 (45%) 0.45 (20%) -0.16 -(7%)
30 Year 5.00 95.47 0.56 (11%) 1.20 (23%) 2.57 (48%) 0.97 (18%)
5.50 97.94 0.72 (17%) 1.12 (26%) 1.93 (44%) 0.59 (14%)
6.00 100.00 0.90 (28%) 0.94 (29%) 1.18 (36%) 0.25 (8%)
6.50 101.58 0.99 (52%) 0.65 (34%) 0.38 (20%) -0.12 -(7%)
Source: Banc of America Securities
14
RMBS Trading Desk Strategy
5/1 REITs If they are whole pool but not exclusively. Provide good yield for a
short duration asset. Many buy higher coupons w/less duration w/the
thought that if the market sells off that dollar prices are more protected.
Also, the profile of the borrower is more short term than other hybrids
implying that market rates would have to move significantly in order for the
borrower to change his financing horizon.
Banks Good short duration asset if it looks attractive vs. their funding.
Money Managers Will play if a good short duration alternative to CMOs and the like.
Hedge Funds
7/1 Banks Good short duration asset if it looks attractive vs. their funding. Not all
banks go out this far but there are some that do.
Money Managers
Insurance Companies
Hedge Funds
Figure 9 shows the effect of instantaneous rises in rates (holding the OAS constant) on the
15
RMBS Trading Desk Strategy
effective duration and the weighted average life (WAL) of close to par-priced hybrid ARMs
and 15-year pass-throughs. As expected, the effective durations of hybrids extend a lot less
than those of 15-years. For example, for a 100 bps increase in the rates, the effective
durations of 3/1 and 5/1 par coupon hybrids extend by 0.5 years and 0.9 years respectively,
while a comparable 15-year pass-through extends by 1.2 years.
In addition to providing extension protection in rates back up scenarios, short reset hybrids
are a good hedge against housing slowdown which in turn can lead to slower turnover
speeds. In Figure 10, we show the impact of 20% and 40% slower housing turnover relative
to our model projections on the prices of some discount hybrids and 15-year fixed rate
securities. To put things into perspective, a 20% and 40% drop in housing turnover relative
to our model projections correspond to a slowdown of 1% - 1.2% and 2.1% - 2.4%
respectively in long term prepayment speeds of 15-years; while, the corresponding numbers
for hybrids range from 0.9% - 1.2% and 1.8% - 2.5% respectively.
If realized turnover speeds are 40% lower than our model turnover speeds, 15-year
discounts will be worth 3+ to 11 ticks less (on an equal OAS basis). In contrast to fixed rate
MBS, short reset hybrids actually appreciate in value at slower turnover speeds. For
example, 4.5% to 5.0% 3/1 hybrids gain 4-5 ticks corresponding to a 40% drop from our
model turnover speeds. The intuition behind this price gain is as follows. As in the case of
fixed rate MBS, the impact of slower turnover speeds on hybrids is negative during their
fixed rate period. However, slower prepayment speeds in the fixed rate period also lead to a
larger pool factor at reset and also implies slower post reset speeds, both of which enhance
the value of the hybrid tail. Overall, the impact of slower turnover speeds is positive for
short reset hybrids. However, the positive impact of slower turnover speeds on a hybrid
value fades as the length of initial reset increases. This is due to combined effect of longer
fixed rate period and reduced tail contribution in longer reset hybrids.
16
RMBS Trading Desk Strategy
In addition, more sophisticated investors also express views on tail valuations through
trades involving seasoned vs. new origination and 5/2/5 vs. 2/2/5 cap structure. However, to
a large extent, success of these trades depends upon the ability to accurately project
outstanding factors at the reset and post reset speeds.
Finally, a large fraction of the ARM investor base invests in Agency as well as Non-
Agency ARMs. These investors employ different tools including price drops, nominal
spread pickup, and OAS pickup to gauge relative value between these two sectors.
17
RMBS Trading Desk Strategy
• For both 3/1 and 5/1 pools, on average, WAC drift tends to be downwards and its
magnitude increases with age6. The increase in the magnitude of downward drift with age
is caused by an increase in cumulative prepayments.
• The standard deviation of WAC drift increases gradually with time. However, in the case
of FNMA 5/1s, the standard deviation increases significantly at WALA 48 months.
• In general, FNMA 5/1s have higher weighted average WAC drift and standard deviation
5
Note that we use gross WAC to carry out drift calculations. Since the servicing fee on the ARM pools is constant, there will be a one to one
correspondence between the gross WAC drift and net WAC (coupon being passed to security holders) drift.
6
Note that the populations across selected WALAs are overlapping but not identical. For example, all WALA 48 pools are part of WALA 36 and
lower WALA populations but the converse is not true. Our conclusion does not change even when we compare the same pools across different
WALAs.
18
RMBS Trading Desk Strategy
than FHLMC 5/1s. However, FNMA/FHLMC differences are less pronounced in case of
3/1 hybrids.
The higher weighted average standard deviation of WAC drift at WALA 48 months in FNMA
5/1s relative to FHLMC 5/1s can be explained by relatively higher loan age dispersion in the
former. Here we measure the weighted average loan age dispersion in a pool as the difference
between the maximum and minimum loan age (see Figure 13). At WALA 48, FNMA 5/1s
have a weighted average loan age dispersion of 16.3 months compared to 5.4 months for
FHLMC. The higher loan age dispersion at WALA 48 means that some of the loans have
already reset in the FNMA pools or are approaching reset (i.e., generally experiencing faster
prepayments). Both these effects could cause the standard deviation of WAC drift to increase.
However, loan age dispersion for newer 0agency production is actually very comparable
between the two agencies.
Figure 12: WAC Drift in Agency Hybrids (Issue WAC – Current WAC)
Another interesting observation regarding WAC drift can be seen in Figure 14, which shows
that WAC drift distribution has a very high skew towards the right. This implies that it is more
likely for the WAC to drift downwards than upwards.
To conclude, empirical data suggest that on an average WAC of hybrids decreases steadily
with time by up to 4 bps to 5 bps over a three to four year period. While the average drift is
quite small, the increasing standard deviation of the drift with time and the presence of right-
19
RMBS Trading Desk Strategy
skewed WAC distributions in ARM pools imply that the risk of receiving lower payments than
the payments indicated by initial coupon on individual pools can be significant. For instance,
using the WAC drift data of FNMA 5/1s from Figure 12, if a hybrid pool is one standard
deviation away from the mean value of WAC drift, its coupon will be lower by 4.4 bps; 7.8
bps; 11.4 bps and 25.7 bps at WALA 12; 24; 36; and 48 respectively from its initial value.
Further, in a normal distribution the probability of being off from the mean by one standard
deviation in one direction is approx. 16%. However, in this case due to a highly skewed
distribution, this probability will be higher. We estimate the loss on the value of the pool to be
6.25 ticks if the coupon drifts in the above manner.
Figure 14: WAC Drift Distribution of FNMA 5/1 Hybrids at Selected WALAs
Balance (Million $)
20000
10000
0
0
−30 −20 −10 0 10 20 30 40 −40 −20 0 20 40 60 80 100
Balance (Million $)
10000
1000
0 4000
7
Creating mega pools reduced our sample sizes significantly across all selected WALAs. However, the reduced sample sizes were still large enough
from a statistical standpoint. For example, in case of WALA 48 FNMA 5/1s, the sample size reduced from 2,441 to 69 when we grouped the pools to
create $100 MM sized mega pools.
20
RMBS Trading Desk Strategy
the standard deviation has reduced from 20.7 bps to 10.5 bps for FNMA pools. Similarly, for
FHLMC pools the standard deviation has reduced from 9.0 bps to 3.1bps.
With these new risk measures if a hybrid pool is one standard deviation away from the mean
value of WAC drift, its coupon will be lower by 3.3 bps; 5.5 bps; 6.5 bps and 15.7 bps at
WALA 12; 24; 36; and 48 respectively from its initial value. In this case, we estimate the
impact on the pool value to be approximately 4 ticks. If we consider a scenario in which
coupon is lower from the mean by two standard deviations throughout, then the impact on the
pool value is 6 ticks. Hence, a 4-6 ticks range represents an upper bound on the impact of WAC
drift based on the empirical evidence. A couple of corollaries of the above result are as follows:
• The maximum pay-up for LA prefix hybrids (FNMA hybrids with a fixed coupon)
should be 4-6 ticks.
• Since most models price hybrids based on a fixed coupon, they would overstate the
price by 4-6 ticks. Assuming that the market prices in WAC drift by demanding wider
spreads, we can attribute 5-7 bps of the OAS pickup that hybrids offer over fixed rates
to this WAC drift.
21
RMBS Trading Desk Strategy
Any analysis or material that is produced by a trading desk has been prepared by a member of the trading desk who supports
underwriting, sales and trading activities.
Trading desk material is provided for information purposes only and is not an offer or a solicitation for the purchase or sale of any
financial instrument. Any decision to purchase or subscribe for securities in any offering must be based solely on existing public
information on such security or the information in the prospectus or other offering document issued in connection with such offering,
and not on this document.
Although information has been obtained from and is based on sources believed to be reliable, we do not guarantee its accuracy, and it
may be incomplete or condensed. All opinions, projections and estimates constitute the judgment of the person providing the
information as of the date communicated by such person and are subject to change without notice. Prices also are subject to change
without notice.
With the exception of disclosure information regarding BofA, materials prepared by its trading desk analysts are based on publicly
available information. Facts and ideas in trading desk materials have not been reviewed by and may not reflect information known to
professionals in other business areas of BofA, including investment banking personnel.
Neither BofA nor any officer or employee of BofA accepts any liability whatsoever for any direct, indirect or consequential damages
or losses arising from any use of this report or its contents.
To our U.K. clients: trading desk material has been produced by and for the primary benefit of a BofA trading desk. As such, we do
not hold out any such research (as defined by U.K. law) as being impartial in relation to the activities of this trading desk.
This report is distributed in the U.S. by Banc of America Securities LLC, member NYSE, NASD and SIPC. This report is distributed
in Europe by Banc of America Securities Limited, a wholly owned subsidiary of Bank of America NA. It is a member of the London
Stock Exchange and is authorized and regulated by the Financial Services Authority.
22