Beruflich Dokumente
Kultur Dokumente
M A R K E T
Q U A N T I T A T I V E
A N A L Y S I S
Mortgages
UNITED STATES
Robert Young
(212) 816-8332
robert.a.young@citigroup.com
New York
This Commentary has been prepared by Markets Quantitative Analysis ("MQA"), which is part of Citigroup Global
Markets' sales and trading operations.
DDDD
M
This Commentary has been prepared by Markets Quantitative Analysis ("MQA"), which is part of Citigroup Global
Markets' sales and trading operations.
November 2, 2005
Contents
I. TBA: Cheapest-to-Deliver
21
29
Acknowledgements
We thank Brett Rose and Lakhbir Hayre for their helpful suggestions, Carsten Schwarting for
guidance on the structure of the specified pool market, and Buchi Ramagopal and Mikhail Teytel
for their insights. The authors would also like to thank Peg Pisani for editing the report and Adela
Carrazana for providing production editing.
November 2, 2005
I. TBA: Cheapest-to-Deliver
The creation of the to-be-announced (TBA) MBS added significant depth to the
mortgage market. By allowing forward contracts with standardized delivery, the
TBA mortgage added the liquidity necessary for the market to flourish. It also gave
the MBS seller the ability to choose which pools to deliver into the forward trade.
TBA trading requires the specification of the agency, maturity term, product, and
coupon, in addition to the par amount, price, and settlement date. However, the
individual pool is not known at the time of the trade, so other pool attributes cannot
be identified.
As prepayment models evolved, market participants realized that certain attributes
were correlated with slower prepayment speeds while others indicated that a pool
had a higher propensity to prepay. Sellers became more discerning in their choice of
what to deliver into a TBA trade, keeping the pools that they deemed to have the
most attractive characteristics. TBA sellers have an incentive to deliver pools with
poor prepayment characteristics: faster than average for premium pools and slower
than average for discount pools. Examples of this adverse selection process in the
TBA MBS market are known as worst-to-deliver or cheapest-to-deliver.
If sellers deliver pools that they believe have poor prepayment profiles into TBA
trades, what do they do when they want to sell pools with good prepayment profiles?
Sellers sell pools with good prepayment profiles on a specified basis. That is, they
sell the individual pool rather than a generic TBA. A specified pool has certain
attributes that predictive models assess as having a likelihood for prepayments that is
different than the overall aggregate average slower for premium coupon pools or
faster for discount coupon pools. Figure 1 shows the outstanding balance for various
categories of specified attributes as of August 2005.
The table lists some of the well-known specified attributes and shows the outstanding
balance for ranges correlated with slower- or faster-than-average prepayment speeds.
While originators have grown more sophisticated in pooling their mortgage loans,
the amount of outstanding balance with characteristics suggesting a pool will have
prepayment speeds that differ significantly from the overall averages is, in general,
limited.
November 2, 2005
Specified Value
Definition
Loan Balance
Outstanding ($MM)
Pct. Total
70,245
40,697
50,235
1,514,467
4
2
3
90
Occupancy Status
Owner Occupied
Second Home
Investor Property
1,512,649
59,622
67,997
90
4
4
Loan Purpose
Purchase
Refinance
614,079
1,026,188
37
61
FICO
Low FICO
Very Low FICO
Normal FICO
38,614
161,857
1,475,174
2
10
88
LTV
High LTV
Very High LTV
Normal LTV
124,261
31,901
1,519,482
7
2
91
Age
WALA <= 3
WALA >= 24
41,720
256,274
1,377,650
2
15
82
WAC
635,479
145,725
1,040,166
38
17,595
1,658,049
1
99
105,610
96,536
139,492
348,695
89,413
895,898
6
6
8
21
5
53
1,675,644
100
Alt-A
Prime Production
GEO
New York
Texas
Florida
California
Illinois
Other States
Grand Total
All Pools
62
November 2, 2005
We will cover the actual valuation techniques later, but for now we will simply state the hierarchy of relative value.
Simply put, higher coupons prepay faster than lower coupons. In other words, 6.00% pools, on average, prepay faster than 5.50%
pools which, in turn, prepay faster than 5.00% pools.
3
Also, the incidence of refinancing for noneconomic reasons soon after closing is similarly low.
Older vintages tend to exhibit slightly slower prepayment speeds as burnout begins to take effect. The composition of borrowers in a
pool changes as more rate-sensitive borrowers self-select out of pools (prepay), leaving less rate-sensitive borrowers to comprise a
growing percentage of the remaining pool balance.
November 2, 2005
ramp, in which prepayment speeds increase linearly for 30 months before leveling
5
off at 6% CPR for the remainder of the mortgage term.
Figure 2. Standard PSA Ramp
8%
7%
PSA Ramp
6%
CPR
5%
4%
3%
2%
1%
0%
0
12
18
24
30
36
42
48
54
60
Month
Source: Citigroup.
After coupon and age, many other attributes can be used to further refine estimates of
prepayment speeds. The market recognizes these attributes and is willing to pay a
premium to TBA for them based on the magnitude and consistency of their
differentials from the aggregate averages.
Originators also recognize the extra value that can be extracted by grouping certain
attributes and pooling their loans accordingly. They take the most valuable loans and
pool them together to capture the markets pay-up for the specific attribute(s). Each
attribute is segregated in a particular order to maximize the total value of the
6
portfolio of loans to be sold. The next few sections cover some of these attributes.
Although the PSA ramp has limited usage in the current market environment, it was the standard expression for prepayment speeds
in the early development of the MBS market. In the current market environment, its main use is related to CMOs (PAC bands).
6
Once all of the attributes that can be isolated to give additional value are segregated from an originators portfolio of loans, the
remainder is packaged into pools that are destined for the TBA market. As shown above, that constitutes the large majority of the
agency mortgage universe.
November 2, 2005
currently available mortgage rate only tells part of the story. Figure 3 shows a
borrowers principal and interest (P&I) payment for a given rate for various loan
balances. The difference in payments represents the reduction available from
refinancing to a lower rate mortgage. For example, for a rate reduction from 7.0% to
6.0%, the monthly P&I payment decreases by $65 on a loan balance of $100,000 but
falls by $197 on a loan balance of $300,000.
Figure 3. P&I Payments Based on Note Rate and Loan Balance
Rate (%)
5.0
5.5
6.0
6.5
7.0
7.5
8.0
$100,000
$200,000
$300,000
537
568
600
632
665
699
734
1,074
1,136
1,199
1,264
1,331
1,398
1,468
1,610
1,703
1,799
1,896
1,996
2,098
2,201
Source: Citigroup.
Specified loan balance pools are marketed according to their maximum loan balance;
7
that is the highest individual loan size in the pool. The generally established categories
are for pools with a maximum loan size less than $85,000 (low loan balance, or
LLB), maximum loan size less than $110,000 (moderate loan balance, or MLB),
8
and maximum loan size less than $150,000 (high loan balance, or HLB).
Originators first pool loans with balances less than $85,000 (LLB) separately. LLB
pools command the highest pay-ups in the world of specified pools. Of the remaining
loans, those with balances less than $110,000 (MLB) are pooled next, followed by
those with a balance less than $150,000 (HLB).
Figure 4 shows prepayment speeds for fully seasoned 2001 production during the
height of the refinance wave in 2003. The S curves from the refinance wave can
only be shown for premium coupon pools where the WAC is at least 50bp in the
money because, as was noted at the time, rates were the lowest they had been in four
decades and the yield curve was close to 300bp steep (ten-year minus two-year
rates). In short, there were no discounts. LLB pools provided tremendous protection
against prepayments, with speeds as much as 30% CPR or more slower than the
aggregate averages for individual cohorts.
Using the average loan size would convey more information about a pool. However, because originators are meticulous about how
they pool loans, most pools have a very narrow range of loan sizes.
8
Some market participants use different nomenclature, referring to the first two categories as LLB1 and LLB2 and the third as MLB,
but the loan size thresholds are consistent for conventional pools. The loan balance thresholds for GNMA pools are $75,000 for LLB,
$110,000 for MLB, and $150,000 for HLB.
November 2, 2005
Figure 4. S Curves for 2001 Production During the Apr 03Sep 03 Refinance Wave
80
70
60
CPR (%)
50
40
30
20
10
LLB
MLB
Total
0
0
50
100
150
200
250
While loan balance is significant for refinances it makes little difference for
9
turnover. When borrowers contemplate moving, they calculate their potential
monthly mortgage payment based on the home they are considering purchasing. If
the payment is too great, they are forced to purchase a less expensive home or
postpone moving. However, the decision to move is essentially independent of the
10
loan balance on the current mortgage. Figure 5 shows the S curves for fully
seasoned 2003 production in the first half of 2005, when rates were still low by
historical standards but higher than the generational lows of 2003, and the yield
curve was relatively flat.
Figure 5. S Curves for 2003 Production During the Current Year, Jan 05Jul 05
CPR (%)
50
45
LLB
40
MLB
35
Total
30
25
20
15
10
5
0
-150
-100
-50
0
50
Rate Incentive (bp)
100
150
200
Discount coupon speeds are mostly due to turnover and, hence, show little difference between size categories.
10
10
The current LTV does determine the amount of equity available for a down payment on a new house.
November 2, 2005
50
40
30
20
New York
Florida
National Average
10
Texas
Puerto Rico
0
0
50
100
150
Rate Incentive (bp)
200
250
11
Puerto Rico pools are actually the most valuable, with prepayment speeds up to 70% slower than the aggregate averages, but the
small outstanding balance makes them hard to find and sparsely traded.
11
November 2, 2005
The states mentioned above were prepaying consistently slower than the national
average for in-the-money pools. The rate incentive only goes down to zero because,
as mentioned, there were no discounts at the time. Figure 7 shows the S curves for
the same states for 2003 production in 2005. Note that Florida is now paying slightly
above the national average.
Figure 7. Slow State S Curves for FNMA 2003 Originations in 2005
70
60
50
New York
Texas
Florida
CPR (%)
Puerto Rico
40
National Average
30
20
10
0
-150
-100
-50
50
100
Rate Incentive (bp)
150
200
250
Collateral from faster-prepaying states such as Illinois and Michigan, which have
low closing costs for refinancing typically ends up as TBA collateral. However, a
discount coupon pool with a high concentration of loans from a state with fast
enough housing turnover speeds could warrant a modest pay-up. Figure 8 shows the
S curves for some historically faster-prepaying states in 2005.
Figure 8. Fast State S Curves for FNMA 2003 Originations in 2005
80
70
60
CPR (%)
50
California
Massachusetts
Illinois
Michigan
National Average
40
30
20
10
0
-150
-100
-50
50
100
Rate Incentive (bp)
150
200
250
12
November 2, 2005
CPR (%)
40
30
20
10
2nd Home
Investor
Owner
0
-100
-50
50
100
Rate Incentive (bp)
150
200
250
12
The agencies charge fees ranging from 1.5% to 2.5%, depending on the LTV.
13
November 2, 2005
The seasoning ramps for the different occupancy status categories have the same
general slopes and are equivalent in length. Figure 10 shows the seasoning ramps for
second homes, investor properties, and owner-occupied properties for pools with an
average loan size greater than $150,000. The investor property ramp is higher than
the owner-occupied ramp, because for pools of the same age (the loans were
originated at the same time), investor property pools have average borrower note
rates that are about 35bp45bp higher. In general, owner-occupied properties merit a
better rate than non-owner-occupied properties. As shown, however, for identical
rate incentives, investor properties prepay more slowly.
Figure 10. Seasoning Ramp by Occupancy Status (Average Loan Size > $150,000)
30
25
CPR (%)
20
15
10
5
2nd Home
Investor
Owner
0
0
12
18
24
30
Months
Sources: CPR and CDR Technologies.
13
The phenomenon is consistent with GNMA prepayment speeds, which were slower than conventional speeds during the refinance
wave of 2003 but recently have been faster than conventional speeds.
14
November 2, 2005
able to refinance into lower-rate mortgages than they qualified for before. Other
borrowers may be able to extract the equity to clear up lingering credit blemishes.
Aggressive solicitation by overstaffed originators has also contributed. With many
prime borrowers already in lower-rate mortgages, originators are willing to reach
down the credit spectrum to refinance less creditworthy borrowers. In the current
environment, those attributes do not command pay-ups on their own.
Market participants look for a combination of attributes that indicate true credit
impairment and may pay-up slightly for pools that they believe include borrowers
who have yet to cure their credit problems. To determine credit impairment, we use a
combination of attributes derived from FNMAs Expanded Approval (EA) program.
The EA program reclassifies loans that were previously rated as caution in
FNMAs Desktop Underwriter (DU) automated underwriting system into additional
14
levels of approval.
Figure 11 shows 2001 production during the 2003 refinance wave. The EA S
curves are considerably below the overall aggregate average, demonstrating the
validity of the credit story at the time. Figure 12 shows 2003 production in 2005. In
this figure, the EA S curves are above the overall aggregate averages, although
they are considerably closer than in 2003.
CPR (%)
EA-1
70
EA-2
60
EA-3
50
Total
40
30
20
10
0
-100
-50
50
100
Rate Incentive (bp)
150
200
250
14
These pools are not identifiable as such, but experience indicates that the average ranges of attributes for EA-1 are a FICO score of
600650, an original LTV of 75%90%, and a servicing spread of 50bp60bp. EA-2 ranges are FICO scores of 590620, LTVs of
80%95%, and a servicing spread of 60bp100bp. EA-3 ranges are FICOs of 560600, LTVs of 80%95%, and a servicing spread of
100bp140bp.
15
November 2, 2005
CPR (%)
EA-1
60
EA-2
50
EA-3
40
Total
30
20
10
0
-100
-50
50
100
150
200
250
Loan Age: Good News for the Young and the Old
The next attribute to be segregated is loan age. Loan age has long been recognized
as a primary determinant of prepayment speeds. The original PSA ramp
institutionalized the fact that speeds on new pools tend to be slower than the overall
average and ramp up over time. Understanding speeds in terms of a ramp is simple
enough for turnover. Borrowers generally do not move within a few months of
buying a new home and taking out a new mortgage. As their tenures in their homes
increase over time, borrowers begin to move and payoff their mortgages.
Refinances do not require borrowers to move, but there are reasons for a ramp to
exist for in-the-money loans as well. The process of taking out a mortgage involves
costs, such as a new appraisal, title search, property recording fees, and the like.
Even if a borrower refinances into a so-called no point mortgage, in which no
origination points are charged for the mortgage, closing costs typically can run up to
several thousand dollars. As a result, borrowers may be reluctant to refinance if they
15
have refinanced very recently, even if rates have declined again.
Pay-ups for loan age are generally small compared with loan balance pay-ups, but
can occur either in the beginning of the ramp or further along the ramp, depending on
whether the pool is in or out of the money. For in-the-money coupons, that means
low-WALA pools; for out-of-the money (discount) coupons, it means older, or fully
seasoned pools. Loan age can command a pay-up in its own right, but it is also used
in conjunction with other attributes. For example, LLB pools command a higher payup if they are also low WALA, as opposed to being a year old.
Of course, loan age is a transient attribute. In other words, a low-WALA pool
remains low-WALA for only a few months. The market is paying up only for in-themoney pools that are three months old or less. A fully seasoned discount can also
command a pay-up, because the expectation is that it will prepay faster than newer
15
Lenders also try to discourage refinances that occur soon after the previous refinance. For example, one major lender states in its
seller guide that an originator that sells it a loan that prepays within 90 days is held responsible for any losses resulting from lost
servicing rights and a premium purchase price.
16
November 2, 2005
pools with the same coupon. Very seasoned pools also have shorter legal final
16
maturities, which can decrease extension risk.
Although the PSA ramp is fixed in length and magnitude, it is useful for
understanding the concept of the seasoning ramp. Modern attempts to model the
seasoning ramp account for the fact that it can vary over time, depending on
economic conditions and the interest rate environment. When the level of interest
rates is high, the ramp tends to be longer and have a gentler slope. As rates decline,
the ramp becomes shorter and steeper, sometimes dramatically so.
Figure 13 shows the seasoning ramp for cuspy coupon pools for three different
periods. In this case, cuspy is defined as having a rate incentive between 0bp and
50bp. During a period of high rates in 2000, the ramp peaked at about 24 months or
so, whereas for the refinance wave of 2003 speeds reached their fully seasoned level
at about 1215 months. In 2005, speeds also reached their fully seasoned level in
about 1215 months and actually begin to taper off slightly after that.
Figure 13. Seasoning Ramp for Cuspy Coupons (0bp50bp Rate Incentive)
60
Apr 00Sep 00
50
Apr 03Sep 03
Jan 05Jul 05
CPR (%)
40
30
20
10
0
0
12
18
24
30
Age (Months)
Sources: CPR and CDR Technologies.
Seasoning ramps can have different lengths depending on the rate incentive, which is
also included in most prepayment models. Figure 14 shows seasoning ramps for five
different levels of rate incentive ranging from 100bp out of the money to 150bp in
the money. For positive rate incentives, the ramp is steeper and peaks sooner. This
illustrates the value of low-WALA pools for in-the-money pools. In addition, for outof-the-money pools the ramp becomes fully seasoned at about 24 months.
16
Very seasoned premium collateral can also be of value owing to favorable convexity characteristics (burnout, for example).
17
November 2, 2005
50
-50bp to 0bp
100bp to 150bp
0bp to 50bp
45
40
CPR (%)
35
30
25
20
15
10
5
0
0
12
Age (Months)
18
24
30
17
From a modeling perspective, this corresponds to the steepest part of the S curve where convexity effects are greatest. Small
changes in the rate incentive have a large impact on prepayment speeds.
18
18
To qualify for TBA good delivery, 25bp is the minimum servicing spread.
November 2, 2005
The additional pay-up is generally restricted to cuspy and slight premium coupons. A
higher-premium coupon has a considerable incentive to refinance regardless of the
servicing spread. Figure 15 shows the effect of different WACs for LLB 6.00% pools
from the beginning of 2003 through August of 2005. The LLB pools with smaller
servicing spreads, corresponding to lower WACs, consistently prepaid slower than
19
LLB pools with higher servicing spreads (higher WACs).
Figure 15. FNMA LLB 6.00% Pools by Servicing Spread, Jan 03Jul 05
LLB - Svcg Spread 25bp50bp
LLB - Svcg Spread 50bp75bp
LLB - Svcg Spread 75bp100bp
Overall Aggregate Average
80
70
60
CPR (%)
50
40
30
20
10
0
Jan 03
Apr 03
Jul 03
Oct 03
Jan 04
Apr 04
Jul 04
Oct 04
Jan 05
Apr 05
Jul 05
19
Even though the higher-servicing-spread LLB prepaid faster than the lower-servicing-spread LLB, they were still considerably
slower than the overall aggregate averages. For example, during the height of the refinance wave in 2003, overall aggregate 6.00%
coupon speeds peaked above 70% CPR compared with the high-servicing-spread LLB speeds of less than 50% CPR.
19
November 2, 2005
CPR (%)
30
25
20
15
10
5
Purchase
Refinance
0
-100
-50
50
100
150
Rate Incentive (bp)
200
250
300
For out-of-the-money pools, refinance loans appear marginally faster than purchase
money loans, lending some credence to the notion that borrower tenure starts further
along the seasoning ramp for refinance loans. However, for in-the-money pools, the
purchase money and refinance curves are virtually on top of each other with
purchases marginally higher on super-premiums.
Figure 17 shows the seasoning ramps for purchase money and refinance loans. The
ramp for out-of-the-money refinances rises faster than the purchase money ramp,
further supporting the notion that borrower tenure is further along the seasoning
ramp for the refinance loans. For in-the-money pools, the ramps for purchase money
and refinance loans cross a couple of times before settling into a long-term pattern
with purchase money loans prepaying slightly faster than refinance loans after about
the first year or so.
Figure 17. Seasoning Ramp by Loan Purpose
35
Out-of-the-Money Purchase
In-the-Money Purchase
Out-of-the-Money Refinance
In-the-Money Refinance
30
CPR (%)
25
20
15
10
5
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30
Age (Months)
Sources: CPR and CDR Technologies.
20
November 2, 2005
Carry-Based Valuation
The traditional method of computing the carry advantage from owning a specified
pool, due to its superior prepayment behavior, is based on dollar roll analysis. The
Bloomberg Roll Analysis calculator compares the difference between holding a pool
and selling it into the roll. The three variables in the roll analysis are the break-even
finance rate, the drop (difference between the front-month and back-month prices),
21
and the prepayment speed.
By using the market rate for the break-even financing rate and varying the
prepayment speed assumption, we can see how the drop changes. In other words, we
can determine the value of the prepayment speed differential between specified pools
and TBA by comparing the drops for the different prepayment speeds.
Figure 18 shows the difference in the drop and the incremental carry advantage for
various prepayment speeds for a FNMA 6% MBS (dollar price 102-04). For
example, a pool that pays at 30% CPR when the aggregate average is 40% CPR
would have a one-tick advantage in carry for that month. Figure 19 shows the
difference in the drop and the incremental carry advantage for various prepayment
speeds for a FNMA 5% MBS (a discount coupon with a dollar price of 98-25).
20
Moreover, the investor is also implicitly short a put option. While borrowers cannot increase their loan balance at the existing note
rate when market rates rise, they can prepay slower than expected. In this case, investors would have to wait longer to receive some of
the principal cash flow than they expected and they would be unable to invest it at the prevailing higher rates. In this sense, relative to
the cash flow that they expected based on their prepayment assumptions when they originally purchased the MBS, they have had a
put exercised against them.
21
For an explanation of dollar rolls, see Dollar Rolls In Practice and Theory, Citigroup, July 1, 2004.
21
November 2, 2005
10
20
30
40
50
60
Back-Month Price
Drop
102-01+
102-02:2
102-03:1
102-04:1
102-05:3
102-06:7
05+
04:6
03:7
02:7
01:5
00:1
00:6
00:7
01
01:2
01+
NA
Back-Month Price
Drop
98-20:7
98-20:6
98-20+
98-20:3
98-20:2
98-20
04:1
04:2
04+
04:5
04:6
05
NA
00:1
00:1
00:1
00:1
00:2
NA Not applicable.
Source: Citigroup.
5
10
15
20
25
30
NA Not applicable.
Source: Citigroup.
The tables show that the carry advantage is larger when the overall rate of
prepayment is faster. For the premium coupon, the carry difference for a 10% CPR
change in prepayment speeds is 1 ticks from 60% CPR down to 50% CPR and 1
ticks from 50% CPR down to 40% CPR.22 Furthermore, the carry advantage is
cumulative, so a specified pool that pays at 40% CPR when the overall aggregate
average is 60% CPR has a 2-tick carry advantage.
Finally, the carry advantage estimated from the roll is the advantage for one month.
Because the advantage is cumulative, each month that the specified pool prepays
slower than the TBA adds to the carry advantage. This makes it possible to estimate
the payback period for a market pay-up. For example, if a specified pool with a
6.00% coupon has a 12-tick pay-up and prepays at 30% CPR when the overall
aggregate average is 40% CPR, the investor will earn the pay-up back in 12
months.23 After that, any advantage in prepayment speeds represents profit on the
trade for the investor, relative to purchasing a TBA.
OAS-Based Valuation
The other main method of valuation is OAS-based valuation. The most common
method for expressing the relative value of specified pools using OAS is to use a
constant OAS as a benchmark to calculate a theoretical price for the specified pool.
The OAS of the TBA is used as a proxy for the specified pool. The difference
22
The carry advantage is also greater for pools that are further from par. For example, if we look at a FNMA 6.50% pool (dollar price
103-08), the carry difference for a 10% CPR drop in prepayments from 50% CPR to 40% CPR is 1 ticks compared with the 1
ticks for the 6.00% pool.
23
22
November 2, 2005
between the theoretical price of the specified pool and the market price of the TBA is
24
the theoretical, or implied, pay-up.
Market pay-ups are typically not as high as theoretical (implied) pay-ups because of
various factors that we discuss below. Figure 20 shows attributes for a generic
FNMA 6.00% TBA pool including the OAS, calculated for a price of 102-04. Figure
21 shows the specified value for various attributes, the theoretical price, and implied
pay-up (using constant OAS). We used the generic TBA attribute values for all
attributes except the specified attribute listed in the table. The table also shows the
weighted average life (WAL), one-year CPR, long-term CPR, the effective duration,
25
effective convexity, the option cost, and the Z-spread for each specified pool.
Figure 20. Generic FNMA 6.00% TBA Attributes
Attribute
Value
Coupon
WAC
WAM
WALA
Loan Balance
FICO
LTV
Servicing Spread
Refinance Pct.
Investor Pct.
OAS (@102-04)
6.00%
6.50%
343 Months
15 Months
$167,000
703
77%
50bp
42%
13%
-13bp
Source: Citigroup.
Figure 21. Theoretical Prices and Implied Pay-Ups for Individual Specified Attributes
Specified
Pool
Specified
Attribute
TBA
LLB
MLB
HLB
Investor
Low FICO
High LTV
High SATO
Low WALA
Fully Seasoneda
Low WAC
Purchase
New York
NA
Avg. Loan Balance
Avg. Loan Balance
Avg. Loan Balance
Inv. Property Pct.
FICO
LTV
Servicing Spread
WALA
WALA
Servicing Spread
Purchase Loan Pct.
State Pct.
Specified
Value
Theoretical
Price
Implied
Pay-Up
WAL
NA
$65,000
$95,000
$125,000
100%
600
90%
100bp
0 Mo.
24 Mo.
25bp
100%
100%
102-04
103-11:2
102-28:7
102-15:5
102-17:2
102-12:1
102-02
102-00
102-04
99-04:6
102-10:4
102-04
102-17:6
NA
39:2
24:7
11:5
13:2
8:1
-2:0
-4:0
0:0
11:6
6:4
0:0
13:6
4.0
4.9
4.6
4.3
4.2
4.0
3.9
3.2
4.2
6.8
4.7
4.0
4.6
One-Year Long-Term
CPR (%)
CPR (%)
29.0
21.7
23.8
25.7
27.8
29.6
28.0
35.7
16.4
13.0
23.1
29.0
14.2
20.9
17.2
18.4
19.4
20.3
21.2
21.8
26.5
20.1
11.3
17.8
20.9
18.5
Effective
Duration
1.88
2.59
2.32
2.07
2.13
2.06
1.91
1.91
2.31
3.68
1.93
1.88
2.60
Effective
Option
Convexity Cost (bp)
-2.78
-2.19
-2.42
-2.65
-2.44
-2.48
-2.70
2.57
-2.79
-1.94
2.91
2.78
-2.61
Z-Spread
(bp)
98
76
84
92
88
90
98
90
105
73
100
98
101
85
63
71
79
75
77
85
77
92
60
87
85
88
a The fully seasoned pool is compared to a FNMA 5.00% TBA with a price of 98-25. NA Not applicable.
Note: New York is run assuming a $227,000 loan size and two-month WALA. See report on manifold MB707.
Source: Citigroup.
24
It is also possible to compare OASs. The TBA price plus the market pay-up is used as the price to calculate an OAS for the
specified pool. The investor can then directly compare the OAS of the TBA and OAS of the specified pool. The OAS of the specified
pool (even with the market pay-up included in the price) is generally higher than the OAS of the TBA because of its superior
prepayment profile. The difference between the OAS of the specified pool and the OAS of the TBA is known as an OAS pickup.
25
Daily specified collateral valuation reports are available on FI Direct: Origination Year Report to Swaps on manifold MB712,
Stipulated Attributes Report on MB721, and Model Payups for State Specific Pools on MB707.
23
November 2, 2005
The most valuable specified pool is LLB, which has an implied pay-up of 39
26
ticks. LLB pools prepay considerably slower as the WAL extends to 4.9 years
(from the TBA value of 4.0 years) and the one-year CPR drops from 29.0% CPR to
21.7% CPR. The duration is about 0.7 year longer and the bond is less negatively
convex.
By comparison, the next most valuable pool is the MLB pool, which has an implied
pay-up of 24 ticks. Its WAL extends to 4.6 years with the one-year CPR at 23.8%
CPR. The other specified pools follow a similar pattern.
The implied pay-ups for high LTV and high SATO are negative, indicating that the
models value these pools at less than the TBA pool.27 The high-LTV and high-SATO
pools have weighted-average lives that are less than TBA, suggesting that they
prepay faster over the life of the pool. Of course, no one would sell a pool for less
than TBA; they would simply deliver the pool into a TBA trade.
The table shows low-WALA pools with no implied pay-up. Low-WALA pools
sometimes command a modest pay-up in the market only for pools with a WALA of
three or less. However, as noted, a low WALA can enhance the value of other
specified attributes. Figure 22 shows implied pay-ups for a few combinations of
specified attributes.
Figure 22. Theoretical Prices and Implied Pay-Ups for Selected Combinations of Specified Attributes
Specified Pool
Specified Attribute
Specified Theoretical
Value
Price
TBA
LLB
LLB w/Low WAC
NA
Avg. Loan Balance
Avg. Loan Balance
Servicing Spread
Avg. Loan Balance
WALA
Avg. Loan Balance
Servicing Spread
WALA
FICO
LTV
Servicing Spread
NA
$65,000
$65,000
25bp
$65,000
0 Mo.
$65,000
25bp
0 Mo.
600
90%
100bp
Credit Impaired
Implied
Pay-Up
WAL
One-Year Long-Term
CPR (%)
CPR (%)
Effective Effective
Duration Convexity
Option
Cost (bp)
Z-Spread
(bp)
102-04
103-11:2
103-20:6
NA
39:2
48:6
4.0
4.9
5.5
29.0
21.7
18.3
20.9
17.2
14.9
1.88
2.59
2.71
-2.78
-2.19
-2.23
98
76
76
85
63
63
103-19:3
47:3
5.5
10.2
15.2
3.17
-1.99
82
69
104-00:7
60:7
6.5
8.4
12.6
3.34
-1.96
80
67
101-31:6
-4:2
2.8
36.5
29.7
2.08
-2.10
83
70
NA Not applicable.
Source: Citigroup.
The table shows that some combinations are worth more than the sum of the
individual attributes. For example, the LLB and low WALA combination has an
implied pay-up of 47 ticks. Why is the total implied pay-up for the combination of
LLB and low WALA worth more than the sum of the individual pay-ups for LLB
and low WALA (39 + 0 = 39 ticks)? Part of the explanation is that Yield Book
recognizes that the seasoning ramps for different loan sizes are different.
26
The theoretical prices and implied pay-ups are listed in trader notation. The units for the numbers before the dash are points, the
numbers after the dash are in ticks (32nds of a point), and the numbers after the colons are eighths of a tick.
27
From the models perspective, the TBA pool has the aggregate average for specific attributes (average loan balance, LTV, FICO,
etc.). As noted, however, credit-related stories that were valid during the refinance wave of 2003 have turned around since then, with
low-FICO and high-LTV pools prepaying faster than the overall aggregate averages.
24
November 2, 2005
Figure 23 shows the seasoning ramps by loan size category in 2005. The ramp for
LLB is below that for the overall aggregate average, graphically showing the
prepayment advantage. Further, the slope for LLB is gentler than the aggregate
average. The differential between the two ramps is increasing and peaks at about
10%12% CPR CPR around 1215 months before narrowing to a long-term
differential of about 5%8% CPR. The difference in the seasoning ramp for LLB
adds extra value beyond what our prepayment model predicts based on the pure loan
balance effect alone. That extra value is recognized when we inform the model of the
28
lower WALA as well.
Figure 23. Seasoning Ramps for In-the-Money Pools (0bp100bp) by Loan Size Category
LLB
35
MLB
All
30
CPR (%)
25
20
15
10
5
0
0
12
18
Age (Months)
24
30
36
28
Simply put, a LLB pool with a WALA of less than 3 is worth more than a LLB with a WALA of 15.
29
For example, a pool cannot simultaneously have a low WAC (with a servicing spread of 25bp) and the high-servicing-spread
characteristic of credit-impaired pools (100bp).
25
November 2, 2005
Figure 24. Theoretical Prices and Implied Pay-ups for a Waterfall of Specified Attributes
Specified
Pool
Specified
Attribute
Specified
Value
Theoretical
Price
Implied
Pay-Up
WAL
One-Year
CPR (%)
Long-Term
CPR (%)
Effective
Duration
Effective
Convexity
TBA
LLB
Investor
Low FICO
Low WAC
Low WALA
High LTV
Best Combination
NA
Avg. Loan Balance
Inv Property Pct.
FICO
Servicing Spread
WALA
LTV
Loan Size, Inv. Prop.
Pct., Low WAC,
Low WALA
NA
$65,000
100%
600
25bp
0 Mo.
90%
NA
102-04
103-11:2
103-16:3
103-15:6
103-25:2
104-02:3
104-07:2
104-11
NA
39:2
44:3
43:6
53:2
62:3
67:2
71:0
4.0
4.9
5.0
4.9
5.5
6.4
6.5
6.9
29.0
21.7
21.0
21.7
18.3
11.4
12.0
7.7
20.9
17.2
16.7
17.2
14.9
12.8
12.4
11.7
1.88
2.59
2.69
2.70
2.82
3.23
3.23
3.49
-2.78
-2.19
-2.08
-2.02
-2.06
-2.00
-2.11
-1.84
Option Z-Spread
Cost (bp)
(bp)
98
76
74
73
73
77
75
76
85
63
61
60
60
64
62
63
NA Not applicable.
Source: Citigroup.
The last line shows the best combination, which is the combination of attributes
that provides the largest implied pay-up. The attributes are LLB and 100% investor
properties, combined with low WALA and low WAC (servicing spread). A low FICO
and/or high LTV actually make prepayments faster, reducing the implied pay-up.
If the implied pay-ups seem too high compared with current market pay-ups, thats
because they are. In general, market pay-ups not are as high as theoretical pay-ups,
because there is uncertainty regarding the recovery of the pay-up. Most importantly,
the prepayment model can be wrong. The difference in speeds predicted by the
model may not materialize.
Beyond model projection errors, market conditions can change to make the
differential in prepayment speeds less than expected or less relevant from a relative
value perspective. The recovery period for a specified pay-up can be many months,
potentially stretching into years. Over this time frame, rates can change, turning a
premium coupon pool into a discount coupon pool, at which point the prepayment
protection can cease to exist.
Also, the roll could become special. A roll is considered special when the drop is
trading above carry. In other words, an investor selling into a dollar roll would pay
less to repurchase the TBA in the back month than indicated by the current market
financing rate and the markets consensus for prepayments. Rolls usually become
special because of technical factors, such as supply shortages or heightened demand
for collateral for CMOs. When rolls are special, the value of specified pools declines
because investors can earn additional carry via the roll. Moreover, because the back
30
month price is determined at the time of the roll, the carry is locked in.
Because the theoretical prices given by the model are often different (usually higher)
compared with market pay-ups, effective durations can differ from observed market
durations. In particular, model LLB effective durations are often substantially longer
than durations implied by market price moves. Investors hedging specified pools
sometimes pick a duration between that of the TBA and the duration given by a
model for the specified pool. For example, if the implied pay-up is 20 ticks and the
market pay-up is 10 ticks, an investor might use a duration that is midway between
30
Investors can earn additional carry via the roll when it is special. For example, if the drop is trading 2 ticks above carry, the
investor is paying 2 ticks less than breakeven for the repurchase in the back month (i.e., earning 2 ticks of additional carry). The
investor must then choose between guaranteed carry for as long as the roll remains special and the carry advantage from specified
prepayment protection, which may last longer but is not guaranteed.
26
November 2, 2005
the model durations for the TBA and specified pool for hedging purposes.
Alternatively, an investor might use an empirical duration, estimating the duration
of the specified pool based on recent rate and price movements in the market.
Carry-Based
OAS-Based
Pricing Assumption
Constant pay-up/price
TBA OAS
Advantages
Simple and intuitive only need nearterm prepayment projections and can
compute breakeven number of
months for earning back specified pool
pay-up.
Disadvantages
Source: Citigroup.
27
November 2, 2005
Figure 26. LLB and TBA 6.00% Pools Constant OAS Immediate Rate Shift Scenario Projected Prices and
Differences, 13 Sep 05 Close
Scenario
FN 6.00% LLB
TBA
Difference
-300bp
-200bp
-100bp
0bp
+100bp
+200bp
+300bp
106-00
103-28
2-04
105-11
103-00
2-11
104-30
102-22
2-08
103-11
102-04
1-07
99-23
99-01
0-23
94-31
94-19
0-12
89-24
89-19
0-05
Source: Citigroup.
The scenario pay-ups are as one might expect. Because LLB provides protection
against refinancings, the relative value of LLB collateral generally increases as rates
decline and refinancings increase. In other words, the bulk of the value of LLB
31
comes from rate rally scenarios.
However, although rallies tend to enhance the value of LLB, the LLB advantage
reaches a maximum in the -200bp scenario. In this scenario, TBA 6.00% pools are so
far in the money that further drops in rates have little impact on projected speeds.32 In
contrast, speeds of LLB 6.00% pools, which are at lower levels than TBA speeds, can
continue to increase as rates decline. (Figure 27 shows projected scenario speeds.) As
a result, the speed advantage of LLB 6.00% pools can be seen to decline as rates
decline from the -200bp to -300bp scenario, and the implied pay-up follows suit.
Figure 27. LLB and TBA 6.00% Pools Immediate Rate Shift Scenario Projected Long-Term Projected CPRs (%),
13 Sep 05 Close
Scenario
FN 6.00% LLB
TBA
Difference
-300bp
-200bp
-100bp
0bp
+100bp
+200bp
+300bp
53.4
72.9
-19.5
45.9
71.3
-25.4
31.3
52.2
-20.9
17.2
20.9
-3.7
10.4
10.4
0.0
8.4
8.5
-0.1
7.6
7.7
-0.1
Source: Citigroup.
31
It follows that an investor who believes rates will increase would tend to value LLB less than others who do not have a view on
rates or expect rates to fall.
32
At this point, speeds have essentially reached their maximum level. See Anatomy of Prepayments The Citigroup Prepayment
Model, Citigroup, March 2004.
28
November 2, 2005
IV. Conclusion
Specified pool trading has long been an important sector of the mortgage market. In
the 1990s, seasoned collateral was actively traded. In recent years, loan size has
tended to dominate specified pool trading, but, as described in this paper, many other
collateral characteristics have been identified as providing added value to investors.
There is good reason to believe future opportunities in specified pools will arise:
Interest in specified pools naturally grew after the agencies increased their pool
level disclosures in 2003. Freddie Mac recently announced that it would begin
providing loan level data for its newly issued pools in the near future. This more
detailed data may provide additional ways in which to identify value; for
example, non-third-party originations and condominiums might exhibit favorable
prepayment behavior.
In recent months, there have not been many dollar roll opportunities. Less value
in rolls translates into greater value in specified pools.
29
Disclaimer
This communication is issued by a member of the sales and trading department of
Citigroup Global Markets Inc. or one of its affiliates (collectively, Citi). Sales and trading
department personnel are not research analysts, and the information in this
communication (Communication) is not intended to constitute research as that term is
defined by applicable regulations. Unless otherwise indicated, any reference to a
research report or research recommendation is not intended to represent the whole
report and is not in itself considered a recommendation or research report. All views,
opinions and estimates expressed in this Communication (i) may change without notice
and (ii) may differ from those views, opinions and estimates held or expressed by Citi or
other Citi personnel.
This Communication is provided for information and discussion purposes only. Unless
otherwise indicated, it does not constitute an offer or solicitation to purchase or sell any
financial instruments or other products and is not intended as an official confirmation of
any transaction. Unless otherwise expressly indicated, this Communication does not
take into account the investment objectives or financial situation of any particular person.
Recipients of this Communication should obtain advice based on their own individual
circumstances from their own tax, financial, legal and other advisors before making an
investment decision, and only make such decisions on the basis of the investor's own
objectives, experience and resources. The information contained in this Communication
is based on generally available information and, although obtained from sources
believed by Citi to be reliable, its accuracy and completeness cannot be assured, and
such information may be incomplete or condensed.
Citi often acts as an issuer of financial instruments and other products, acts as a market
maker and trades as principal in many different financial instruments and other products,
and can be expected to perform or seek to perform investment banking and other
services for the issuer of such financial instruments or other products.
The author of this Communication may have discussed the information contained therein
with others within or outside Citi and the author and/or such other Citi personnel may
have already acted on the basis of this information (including by trading for Citi's
proprietary accounts or communicating the information contained herein to other
customers of Citi). Citi, Citi's personnel (including those with whom the author may have
consulted in the preparation of this communication), and other customers of Citi may be
long or short the financial instruments or other products referred to in this
Communication, may have acquired such positions at prices and market conditions that
are no longer available, and may have interests different from or adverse to your
interests.
Investments in financial instruments or other products carry significant risk, including the
possible loss of the principal amount invested. Financial instruments or other products
denominated in a foreign currency are subject to exchange rate fluctuations, which may
have an adverse effect on the price or value of an investment in such products. No
liability is accepted by Citi for any loss (whether direct, indirect or consequential) that
may arise from any use of the information contained in or derived from this
Communication.
Past performance is not a guarantee or indication of future results. Any prices provided
in this Communication (other than those that are identified as being historical) are
indicative only and do not represent firm quotes as to either price or size. You should
contact your local representative directly if you are interested in buying or selling any
financial instrument or other product or pursuing any trading strategy that may be
mentioned in this Communication.
Although Citibank, N.A. (together with its subsidiaries and branches worldwide,
"Citibank") is an affiliate of Citi, you should be aware that none of the financial
instruments or other products mentioned in this Communication (unless expressly stated
otherwise) are (i) insured by the Federal Deposit Insurance Corporation or any other
governmental authority, or (ii) deposits or other obligations of, or guaranteed by, Citibank
or any other insured depository institution.
IRS Circular 230 Disclosure: Citi and its employees are not in the business of
providing, and do not provide, tax or legal advice to any taxpayer outside of Citi. Any
statements in this Communication to tax matters were not intended or written to be used,
and cannot be used or relied upon, by any taxpayer for the purpose of avoiding tax
penalties. Any such taxpayer should seek advice based on the taxpayers particular
circumstances from an independent tax advisor.
2011 Citigroup Global Markets Inc. Member SIPC. All rights reserved. Citi and Arc
Design are trademarks and service marks of Citigroup Inc. or its affiliates and are used
and registered throughout the world.