merrill lynch - valuing subordinate abs
DESCRIPTION
Introducing a New Risk-Adjusted Framework Based on Home PricesTRANSCRIPT
-
Highlights
The role of home prices in the valuation of mortgage-backed securities has long
taken a back seat to that of interest rates. However, over the past few years, investors
have become increasingly aware of the significant impact of home prices on
performance, both in terms of prepayments and in terms of credit. This is most
pronounced in the sub-prime ABS market, where home prices may be the most
dominant factor in driving future performance, especially for subordinate securities.
This relationship has been brought to the forefront over the past several months, as
investors have sought to purchase protection in the ABS credit default swap (CDS)
market, against a downturn in the housing market. Yet, despite this flurry of activity
and exceptionally volatile spreads, there has been little consensus on the appropriate
approach to valuing these securities. In this paper, we present an innovative, option-
based approach to pricing this risk and understanding the implications of various
spread levels. In particular, we find the following:
Home price appreciation going forward may be weaker than it has been over the
past several years. Although hard evidence of a slowdown has been limited, the
most recent data seem to indicate that home price appreciation has decelerated.
Sub-prime mortgages are particularly sensitive to the rate of home price
appreciation (HPA). The rate of HPA affects prepayments, defaults, and loss
severities. Taken together, these findings imply that cumulative losses are
extremely dependent on the housing market.
Armed with the dependence of prepayments and losses on home prices, we are
able to calculate a price for each security in an ABS structure for any given
home price scenario. This immediately allows us to judge what home price
scenarios are required for an ABS subordinate to begin losing value.
Rather than choosing one home price appreciation rate, it is more reasonable to
assume a distribution of possible scenarios. We find that different home price
distributions correspond directly to various spread levels. Not only is the mean
of the distribution important, but also its standard deviation (or volatility). We
accordingly treat subordinate ABS as (short) options on home prices.
By reversing the approach and using observed market spreads to derive an
implied HPA distribution, we establish a calibration with which we can price the
HPA risk embedded in a range of securities. It also provides direct insight into
capital structure pricing and potential relative value opportunities.
In the process, we create a ABS valuation framework that incorporates home prices,
interest rates, and deal structure in a single option adjusted framework.
ASSET-BACKED SECURITIES
Contributors
Kamal Abdullah MBS Strategist, MLPF&S (1) 212 449 9308 [email protected]
Akiva Dickstein MBS Strategist, MLPF&S 1) 212 449 1759 [email protected]
Shaolin Li ABS Strategist, MLPF&S 1) 212 449 6891
Sarbashis Ghosh ABS Strategist, MLPF&S 1) 212 449 4457 [email protected]
Jonathan Braus MBS Strategist, MLPF&S 1) 212 449 9728
3 February 2006
Valuing Subordinate ABS Introducing a New Risk-Adjusted Framework Based on
Home Prices
Merrill Lynch does and seeks to do business with companies covered in its research reports. As a result, investors should be aware that the firm may have a conflict of interest that could affect the objectivity of this report.
Investors should consider this report as only a single factor in making their investment decision.
Refer to important disclosures on page 25. Analyst Certification on page 24.
Global Securities Research & Economics Group Fixed Income StrategyRC#41403401
-
Valuing Subordinate ABS 3 February 2006
2 Refer to important disclosures on page 25.
CONTENTS
Section Page
Section I Introduction 3
Section II Todays Housing Market 5
Section III Home Prices and the Sub-prime Borrower 8
Section IV Pricing Housing Market Risk Credit-Adjusted OAS 12
Section V Final Thoughts and Direction for Future Work 21
-
Valuing Subordinate ABS 3 February 2006
Refer to important disclosures on page 25. 3
1. Introduction
Over the past several months ABS investors have been grappling with
valuation issues, as spreads on subordinate securities (Baa3, for example)
widened by up to 200 bp in October before tightening back by more than 100 bp
toward the end of 2005 and beginning of 2006. Two factors precipitated much of
this widening: i) an increasing number of investors wanting to take positions
against the housing market and ii) the belief that ABS CDS1,2 (Asset-Backed
Securities Credit Default Swaps) was the best way to do so.
Although spreads have moved dramatically over the past few months, there
remains little consensus as to their proper level. Is a spread of 200 bp rich for
Baa3 ABS? Is 400 bp cheap? What is the appropriate spread level required
to fairly capture the risk inherent in these securities?
This paper will develop a methodology for answering these questions. Our goal
is to understand precisely what kind of housing market is implied by a
particular level of spreads so that we can make informed investment
decisions.
We begin by briefly reviewing housing market trends and show that recent data
have finally indicated that the housing market may be softening. We also discuss
some of the factors that may weigh on home price appreciation going forward,
including declining affordability and a negative media effect.
We then address the impact of home price appreciation (HPA) on sub-prime
borrowers and present a historically motivated model of prepayments, defaults,
and loss severities as a function of home prices. We show that home price
scenarios play a major role in driving all three of these variables, and thereby
serve as perhaps the most critical economic determinant of cumulative losses.
Given that subordinate ABS performance is driven by cumulative losses, and these
losses are inherently driven by home prices, then valuations should depend
directly on home price appreciation.3 Consequently, we can calculate a fair
price on each subordinate security for any given home price scenario.
Of course, it is difficult to project a single home price scenario with any degree of
certainty. Rather than project a single HPA scenario, we look at value across a
probabilistic distribution. We can take any home price appreciation distribution
and calculate the associated fair value spreads on subordinate securities. These
spreads can then be compared with market spreads to assess whether ABS
subordinates are attractive or expensive. Essentially, this approach represents the
introduction of an option-adjusted framework that puts home prices on near equal
footing with interest rates.
We can also reverse this approach: rather than choosing a home price
distribution and calculating fair value spreads, we can take market spreads
and calculate an implied distribution of home price scenarios. This framework
is analogous to the familiar option-based OAS approach used throughout fixed
income. Just as OAS employs a market-implied price for interest rate risk, this
approach does the same for HPA via the burgeoning market for ABS CDS. As
1 An ABS CDS swap is a synthetic contract between a protection buyer (shorting bonds) and seller
which is modeled after the eponymous, successful corporate contract. Sector and cashflow differences
have given rise to a PAYGO or PAUG structure which is better suited to ABS cashflows. There
are minor cashflow differences between ABS CDS and their cash equivalents including interest
shortfall payments. We ignore these minor differences here. For more information see Lang Gibson,
Structured Finance CDS & Implications for the CDO Market, April 27, 2005, Merrill Lynch
Research. 2 Most recently, trading has begun in ABX.HE a set of synthetic ABS indices that enable one to trade
baskets of single-name ABS CDS. This development further facilitates our proposed methodology by
creating generic and transparent spreads for different points in the capital structure. 3 ABS prices also depend on interest rates which are usually correlated with home prices. Deal
structure, on the other hand, is fully determined at issuance it is not a driver of valuation, but rather a
predictable translation table from drivers to bond cashflows.
Home prices have a major
impact on prepays, defaults and
severities, and thus are a
dominant driver of cumulative
losses.
We can calculate the fair price
for any given HPA scenario.
Introducing an option-adjusted
framework: valuing ABS using
a distribution of home price
scenarios.
Market spreads can also be
used to imply a distribution of
home prices.
-
Valuing Subordinate ABS 3 February 2006
4 Refer to important disclosures on page 25.
swaps and swaptions imply forward rates and volatilities, market spreads on
mezzanine and subordinate ABS, by analogy, imply forward home prices and their
volatility. This implied distribution can be used to value other ABS and generate
rich/cheap relationships across the capital structure.
-
Valuing Subordinate ABS 3 February 2006
Refer to important disclosures on page 25. 5
2. Todays Housing Market
A Spectacular Run for Home Prices, Especially in Select Regions
Over the past decade, U.S. homeowners have enjoyed near continuous and, at
times, spectacular levels of home price appreciation. Geography has played a
determining role. We summarize home price appreciation by state according to
cumulative five-year HPA using the Freddie Mac/OFHEO repeat sales index
(Chart 1).4 The most recent HPA wave which started in 2004 has been even
more pronounced, with some states appreciating in excess of 25% per annum
while others have remained below 5% (shown in the percentages in Chart 1).
Chart 1: Home Price Appreciation by State
Source: OFHEO
Is the Housing Market Slowing?
Reasons that the Housing Market May Slow
There are a number of reasons to suspect that the recent run in housing prices may
slow over the next few years. First, affordability is down (Chart 2).5 The
composite and first-time homebuyer indices were 138.4 and 80.9 respectively
during 2003. These readings have deteriorated to 117.8 (a 15% decline) and 68.4
(a 16% decline) as of 3Q2005 a direct result of higher home prices and interest
rates.
4 There are numerous methods used to gauge home price growth, each with its own idiosyncrasies and
shortcomings. For the purpose at hand, we simply employ the widely used OFHEO repeat sales index
which is calculated using selected properties that back agency conforming loans. 5 An affordability reading of 100 means that a family with the median income (U.S. Census) exactly
qualifies for a mortgage loan at the national average blended arm-fixed rate on a home valued the
median home price (FHFB).
Five-Year Cumulative HPA
Red States 20-30%
Yellow States 30-60%
Blue States 60+%
Most Recent
1-Year HPA
19%
12%
30% 25% 5%
12%
-
Valuing Subordinate ABS 3 February 2006
6 Refer to important disclosures on page 25.
Chart 2: NAR Affordability Index: Declining Affordability
0
20
40
60
80
100
120
140
160
Mar-81
Mar-83
Mar-85
Mar-87
Mar-89
Mar-91
Mar-93
Mar-95
Mar-97
Mar-99
Mar-01
Mar-03
Mar-05
Index
Composite
First-Time Buyer
Source: NAR
A second reason that housing could slow going forward is that the perception that
housing could slow going forward a perception that is growing in popularity. As
with any other asset, todays transaction prices are significantly shaped by buyer
and seller expectations of future prices. As sellers anticipate weaker prices ahead,
they are more inclined to negotiate. Buyers, similarly, are less likely to pay up. In
fact, a prevailing belief that the housing market is cooling undermines many a
buyers foundation that housing is a good investment. New buyers in particular
may feel that they can afford to wait, whereas in the past, many felt that if they did
not buy immediately, prices would only increase further. Most housing market
news presently is dedicated to questioning the housing markets sustainable
growth. For this reason alone, it might not be sustainable.
Some Hard Evidence of a Potential Slowing Has Finally Emerged
We have believed that factors such as affordability and market perceptions could
slow the housing market, but hard evidence was near absent throughout most of
2005. However, recent data suggest that the market may in fact be weaker than it
was several months ago.
For example, according to the NAR, existing single-family supply has been
climbing slowly throughout 2005 from 3.8 months (January) to 4.9 months
(November) almost reaching the indexs 10-year average value of 5.0 months.
Perhaps more importantly, home prices have actually given up ground over the
past few months, even when seasonally adjusted (Chart 3). Over the past six
months, the seasonally adjusted annualized rate of growth is now 5.4%, compared
with 8% for the prior six month period.
Could a negative media effect
also play a role in slowing the
housing market?
Hard evidence of a slowdown
has been limited but may have
recently emerged.
-
Valuing Subordinate ABS 3 February 2006
Refer to important disclosures on page 25. 7
Chart 3: Even Seasonally Adjusted, Home Price Growth Has Flattened
Source: NAR, Merrill Lynch
-
Valuing Subordinate ABS 3 February 2006
8 Refer to important disclosures on page 25.
3. Home Prices and the Sub-Prime Borrower The movement in ABS spreads over the past few months can be tied to investor
indecision about the future direction of home prices. The volatility of the sector
also indicates that investors may be equally uncertain about the appropriate level
of spreads for the associated risk. The goal of this paper is to address this question.
In order to do so, we first must clarify the connection between home prices and
collateral performance.
The sub-prime sector is perhaps the only arena in the mortgage backed securities
market in which home price exposure dominates that of interest rates. On the
prepayment side, lower credit borrowers display significant sensitivity to home
prices. They generally have fewer financial resources beyond their home and have
come to rely on its appreciation to fund other financial obligations through cash-
out refinances. The recent strong housing market accompanied by agreeable
interest rates led sub-prime borrowers to extract home equity through cash-out
refinancings at a record pace in 2004. Deals backed by 2003 vintage loans
routinely have been prepaying at 40-50 CPR (or more) during the past year,
despite no obvious rate incentive. On the credit side, strong home prices have
similarly kept defaults low, as borrowers who could no longer afford monthly
payments were more likely to sell the home and pocket any built up equity rather
than default on the loan and relinquish the property.
We are interested in quantifying the relationship between prepayments and
defaults on one hand and home prices on the other. It is impossible to do this on a
national level because the housing market has been too strong. Simply put, the
sub-prime market has yet to be proven in a nationally weak housing market. This
is very fortunate for borrowers and investors, but somewhat unfortunate for
researchers!
This relationship, though, can still be addressed through the use of a horizontal
analysis in which HPA variations by geography correlated with local loan
performance serve as a proxy. For example, if two pools with identical
characteristics (FICO, LTV, grade ) were located in Texas and California, the
performance of each would be tied to local HPA. If Texas and California pools
prepay at 25 CPR and 50 CPR respectively, the disparity would be attributed to
differences in state level HPA (say 4% and 20% per annum).6
We utilize this approach using Metropolitan Statistical Area (MSA) level home
prices and performance. We provide historical sub-prime prepayments and
defaults by HPA level (Charts 4a & 4b). We have included all securitized sub-
prime loans originated between 1999 and 2005.7,8
6 This approach has two shortcomings. First, other relevant variables at the geographic level are
subsumed in to the HPA variable (unemployment, etc). Second, loan performance in a national
housing downturn could be unlike that observed locally. Nonetheless, the horizontal analysis of HPA is
the only empirically available method of extracting home price driven performance. 7 The average LTV = 80%, the average FICO = 618, 75% of the loans have prepayment penalties, and
the product mix is 25% FRM / 55% 2/28 Hybrids / 20% Other. The observation period is 1999-2005. 8 The additional structure visible is due to inclusion of all loan types (fixed, 2/28s, 3/27s, all penalty
types.)
The sub-prime sector is
especially sensitive to the
housing market.
Use a horizontal analysis to
tie HPA to loan performance
With home prices mainly rising,
how do you quantitatively
determine their impact?
-
Valuing Subordinate ABS 3 February 2006
Refer to important disclosures on page 25. 9
Chart 4a: Historical Subprime Prepayments by Age and HPA
0
10
20
30
40
50
60
70
80
3 6 9 12 15 18 21 24 30 36 48 60
Age (months)
CPR
< 2.5%
5%
10%
15%
20%
Source: Merrill Lynch and Loan Performance
Chart 4b: Historical Subprime Defaults by Age and HPA
0
2
4
6
8
10
12
14
16
3 6 9 12 15 18 21 24 30 36 48 60
Age (months)
CDR
< 2.5%
5%
10%
15%
20%
Defaults are defined as 90+ days delinquent Source: Merrill Lynch and Loan Performance
The sensitivity of sub-prime loan performance to home prices is astounding. A
pool exposed to 15% HPA versus flat home prices prepays 20 CPR faster.
Similarly, by month 36, the same pool would be defaulting at only 8 CDR instead
of 12 CDR.
An interesting observation involves the < 2.5% HPA and 5% HPA buckets.
Although both prepay at nearly the same speeds, they exhibit marked default
differences with age. This makes sense because there is a financial threshold
below which cashout refinances are not feasible. Nevertheless, just 5% equity
growth per year materially reduces default likelihood.
Loss severity, like prepayments and defaults, is a crucial ingredient when
accounting for deal performance. As can be seen, HPA profoundly shapes it as
well (Chart 5).9
9 The observed loss severity seasoning curve can be explained for newer loans by
Subprime performance goes
hand-in-hand with home price
appreciation.
High HPA low loss severity.
-
Valuing Subordinate ABS 3 February 2006
10 Refer to important disclosures on page 25.
Chart 5: Loss Severity versus Home Price Appreciation
-
5
10
15
20
25
30
35
40
45
50
3 6 9 12 15 18 21 24 30 36 48 60
Age (months)
Lo
ss
Se
ve
rit
y (
%)
HPA < 2.5%
HPA=5%
HPA=10%
HPA=15%
HPA=20%
Source: Merrill Lynch and Loan Performance
The are several reasons for this dependence of severity on HPA:
Resale environment recovered properties are more easily resold in a strong
market, reducing upkeep and servicer corporate advances.
Process time the period from foreclosure to resale can typically require
anywhere from 6 to 18 months. Any home price appreciation during that
period applies directly to the homes value and diminishes severity
accordingly. In high HPA environments (e.g., recent California experience),
this can give rise to near zero severities.
Resolution method how the property is transferred and ultimately sold is
also influenced by home prices. Less costly methods such as a short sale or
deed-in-lieu of foreclosure make more sense to both borrowers and lenders in
a strong housing environment.
These factors prepayments, defaults, and severities interact and nonlinearly
contribute to cumulative loss. As higher HPA accelerates voluntary prepayments,
loans that would otherwise fail are able to refinance / payoff and thus make the
investor whole. Fewer loans default anyway and those that do experience lower loss
severities. We schematically show this compounded HPA sensitivity (Chart 6).
The underlying property is still near appraisal value.
The property has spent limited time on its way to foreclosure.
Total advances are lower.
Faster prepayments arising
from high HPA also minimize
cumulative losses
-
Valuing Subordinate ABS 3 February 2006
Refer to important disclosures on page 25. 11
Chart 6: A Schematic of Deal Losses and Home Prices
High HPA
Faster Prepayments
Fewer Defaults
Lower Losses
Deal Structure
Lower Loss
Severity
Source: Merrill Lynch
We provide historical cumulative sub-prime losses by HPA and loan age in Chart
7. The underlying data are the same as those in Charts 4a and 4b.
Chart 7: Cumulative Loss versus Age by Home Price Appreciation
0
1
2
3
4
5
6
7
3 6 9 12 15 18 21 24 30 36 48 60 82
Age (months)
Cu
m L
os
s (
%)
HPA=1.5%
HPA=5%
HPA=10%
HPA=15%
HPA=20%
Source: Merrill Lynch and Loan Performance
Investors concerned about
losses should focus first on the
housing market.
-
Valuing Subordinate ABS 3 February 2006
12 Refer to important disclosures on page 25.
4. Pricing Housing Market Risk Credit Adjusted OAS (COAS)
Relating HPA Scenarios to ABS Spreads
The analysis in the previous section revealed that housing price appreciation
plays a major role in determining sub-prime cumulative losses prepayments,
and enabled us to quantify the relationship between the two. The next key
step is to relate these findings to an understanding of ABS spreads.
Because ABS prepayments and losses are directly related to home price
appreciation, there is validity in the concept of buying protection on ABS CDS as
a way of making a bet that home price appreciation will decelerate. However,
buying protection (shorting a bond) has a cost: the spread on the security must be
paid. In the Treasury market, at a given price, even bearish investors might be
induced to own a security for its yield. In the mortgage market even those who
believe that prepayments are increasingly efficient will take the convexity risk at
the right price. The same should be true for a sub-prime subordinate bond and
home price risk. Given any prior beliefs about the direction of the housing market,
there is some price or spread where a bonds expected value is acceptable. The
question before us, then, is how to value a subordinate security given its
exposure to home prices.
In one sense, this task is far more challenging than determining the impact of
interest rates on MBS. Interest rate risk has long been tamed quantitatively via
familiar financial tools such as term-structure models and OAS, both of which are
predicated on the existence of a liquid interest rate derivatives market. By contrast,
there has been no such market for home prices,10 though the Chicago Mercantile
Exchange will begin trading home price futures as of 2Q06.11
Our approach is three-pronged:
First, we will use the information in the previous section to value ABS
securities under any given home price scenario. This allows investors with
a firm view of future home prices to choose bonds appropriately.
Second, we will use a distribution of these home price scenarios to price
the ABS securities. This is a risk-based pricing approach that allows us to
balance the spread and the risk on these securities. We will show how
choosing the mean and standard deviation of the distribution plays a major
role in subordinate valuation, and discuss the concept of viewing subordinate
ABS as options struck on home prices.
Finally, we will examine the concept of using the ABS CDS market to
imply a distribution of home prices. In purchasing protection, an investor is
buying insurance. The spread paid is therefore the price of risk. We can find
the distribution of home prices which best fits the observed pricing of
subordinate CDS. We can then use this distribution to derive a rich/cheap
relationship across the credit spectrum.
A Reference Deal
In order to be concrete, we choose a reference deal that is representative of recent
sub-prime production in terms of collateral, structure, originator-servicer
reputation, and ratings coverage. Ameriquest 2005-R5 is triple rated by Moodys,
10 As such, housing market risk has not been diversifiable a risk that is readily measured and is
explicitly hedgable. In economics, an investor is not compensated for taking such risk. To illustrate, a
Agency CMO buyer might compare bonds on an OAS basis and then use swaps to dynamically hedge
the purchased bond. A subprime BBB buyer, on the other hand, has had no simple way to hedge home
price exposure and accordingly could not price the risk. 11 http://www.cme.com/trading/prd/env/housingover16250.html
We now turn to the critical
question: How do we relate
HPA to spreads?
Three approaches to the
valuation question.
ABS CDS are indirect bets on
home prices their spreads
indicate implied home price
expectations.
-
Valuing Subordinate ABS 3 February 2006
Refer to important disclosures on page 25. 13
S&P, and Fitch for all classes in the deal. We detail the lower mezzanine structure
and collateral backing it in Tables 1a & 1b.
Table 1a: Ameriquest 2005-R5 Mezzanine Structure
Class Moodys Coupon (bp) Cash (bp) ABS CDS (bp) Initial Size C/E
Senior Classes
M6 A3 70 80 80 1.26% 4.70%
M7 Baa1 122 140 135 1.01% 3.70%
M8 Baa2 135 185 190 0.96% 2.75%
M9 Baa3 175 300 300 0.55% 2.20%
M10 Ba1 300 725 765 0.55% 1.65%
Subordinate Classes
Market spreads as of 11/24/05 Source: Ameriquest, INTEX, and Merrill Lynch
Table 1b: Ameriquest 2005-R5 Collateral
ARM / Fixed 81% / 19%
Owner Occ 98%
SF / PUD 90%
Avg FICO 613
Orig LTV 77%
Full Doc 75%
IO 9%
CA 15%
Avg Loan Size $150k
Source: Ameriquest, Bloomberg
We restrict our attention to the lower mezzanine (A3 and below) capital structure
because the performance of these classes is most sensitive to home price
appreciation in the region of common interest. Home prices would have to drop
dramatically in order for the A1A (Aaa rated) class to incur a loss a possible,
albeit very unlikely, event. The lower mezzanine classes, on the other hand, are
more sensitive to future home prices.
Having chosen our example transaction, we are now ready to discuss our three
approaches to valuation.
Capital Structure Valuation Under Static Pricing
Our first approach (which will also be used as a stepping stone for the later
approaches) is to value each security under the spectrum of possible HPA
outcomes. To do this, we use the collateral models from the previous section;
specify the HPA scenario and interest rate assumptions along with a discount
margin to statically price a given bond.12
12 Its imperative that one not confuse the various components involved in pricing. The external drivers (interest rates and home prices) represent the risks. The models (prepayment, default, delinquency, and
loss severity) are predetermined functions that map rates and home prices into collateral cashflows.
Finally, the deal structure (fully determined and available via INTEX) allocates these cashflows among
the bonds contingent on rates and collateral performance.
Focus on lower mezzanine
classes.
-
Valuing Subordinate ABS 3 February 2006
14 Refer to important disclosures on page 25.
Chart 8: Pricing using HPA, Interest Rates, and a Discount Margin
Prepayments
Defaults
Delinquencies
Loss Severity
Deal
Structure Price
HPA
Int Rates
DM
Source: Merrill Lynch
Throughout this approach, we discount cash flows at LIBOR flat (discount
margin = 0). At first glance, this might seem surprising given that subordinate
ABS offer significant margins. However, if we were to use the bonds stated
margin, we would find that in scenarios where the bond does not incur a loss, the
price is todays market price, but the price is significantly lower at times when the
bond incurs a loss. In other words, prices would have nowhere to go but down.
We believe this is not only awkward but also misses the point: subordinate ABS
investors earn handsome margins by accepting the risk of losses. In scenarios
without losses, the value of a L+300 security should be greater than its initial
market price in order to balance those in less favorable scenarios.13
In Chart 9, we calculate the prices of the various securities in the capital structure
as a function of HPA assuming a fixed LIBOR of 4.5%.
Chart 9: Bond Price versus HPA for fixed LIBOR
0
20
40
60
80
100
120
140
-15% -7% -3% 1% 5% 9% 13%
HPA
Bond P
rice
M6 (A3)
M7 (Baa1)
M8 (Baa2)
M9 (Baa3)
M10 (Ba1)
M11 (Ba2)
Source: Merrill Lynch
13 This approach is also analogous to interest rate derivatives, in which a term structure model is adjusted so that each security has 0 OAS.
How does the capital structure
stand up to home prices?
-
Valuing Subordinate ABS 3 February 2006
Refer to important disclosures on page 25. 15
Chart 9 reveals the breaking point for each bond in the capital structure. The
more complex features arise from the interaction of the collateral model with the
deals structure.14
This simple approach significantly extends the usual first-loss coverage multiple
analysis. Rather than calculating just a single multiple of the default model at
which a bond incurs its first dollar of principal loss, we can size up precisely what
levels of HPA lead to bond price declines and principal losses. The impact of
structure is readily seen.
For example, if one anticipates flat home prices (HPA=0%), an investment
portfolio containing Baa2 and above credit is recommended.
Toward a Risk-Based Pricing Methodology
While Chart 9 gives us the value of each security in a given home price scenario, it
makes more sense to value a security across a range of possible home price and
interest rate scenarios, just as we value interest rate options or mortgages using a
distribution of interest rate scenarios.
To do this, we combine the calculation of the bonds price under each scenario
(we did this in the previous section) with a probability of such a scenario
occurring. The option-adjusted price of the bond will then be the probability-
weighted average of the prices across the various scenarios. We therefore require:
a prescribed interest rate distribution
a prescribed home price appreciation distribution
a correlation between these two15
We present a schematic of the option adjusted pricing method in Chart 10.
Chart 10: Option Adjusted Price From Rate and HPA distributions
Forward interest rate probability distribution (already known from usual term structure)
Correlation
Each scenario uses modeled cashflows
Option Adjusted Price
Forward HPA probability distribution Either we specify or infer from ABS CDS / ABX.HE
Source: Merrill Lynch
14 With lower HPA, losses increase and prepayments decline causing bonds to extend and rely on overcollateralization when available. Ultimately, under sufficiently weak HPA, write-downs ultimately
extinguish the bonds. 15 Home prices and rates may be inversely correlated, at least locally. As rates rise, homes become less affordable and demand accordingly wanes. As rates fall, demand increases. This correlation is relevant
because typical deal structures involve moving parts that are codependent on rates and collateral
prepayments and defaults, both of which depend on home prices. Because home prices are a direct
input in this approach (instead of estimating them from interest rates using the correlation), correlation
specification is less critical.
Dont go below Baa2 if you
expect flat home prices.
Introduce an option framework
to price risk.
Calculate option adjusted price
from HPA and interest rate
probability distributions.
-
Valuing Subordinate ABS 3 February 2006
16 Refer to important disclosures on page 25.
One of the two risk dimensions in Chart 10 is already known. Forward LIBOR and
its volatility (probability shape) are well determined by the familiar capital
markets instruments and can be applied directly. The home price distribution can
either be specified a priori or inferred from ABS CDS (next section).
Investor Beliefs and Pricing
We now consider two hypothetical investors: a housing Bull and a housing
Bear. The Bear anticipates a distribution of housing prices centered around 4%
HPA per year, while the Bull more optimistically predicts a distribution centered
around 11% HPA per year a continuation of recent trends. Both claim a
reasonably modest uncertainty of 7%. Chart 11 compares HPA expectations.
Chart 11: Housing Bear and Bull HPA Outlook
Expected Home Price Appreciation
0%
1%
2%
3%
4%
5%
6%
7%
-15 -5 5 15 25
HPA (%)
Probability
Bullish HPA
Bearish HPA
Source: Merrill Lynch
How would these two interpret the actual market portrayed in Table 1a? Given
their different projected home price distributions, each would calculate a different
fair value price for the securities. In other words, while the two investors may
agree on the price of the security under any given home price scenario, they
weight those scenarios quite differently and as a result arrive at very different fair
value (or option-adjusted) prices. We can translate these fair value prices into a
breakeven or fair value spread at which each investor finds the market cash
prices fair. We show the actual spread along with the Bears and Bulls
breakevens in Table 2.
Table 2: Housing Bear and Bull Breakeven Spreads (Volatility=7%)16
Class Moodys Market Cash Price $
Market ABS CDS Spread (bp)
Bull B/E Spread (bp)
Bear B/E Spread (bp)
M6 A3 99-21 80 10 137
M7 Baa1 99-19 135 24 236
M8 Baa2 98-04 190 82 481
M9 Baa3 95-25 300 184 778
M10 Ba1 86-02 765 359 1170
Rich Cheap
Source: Merrill Lynch
16 as of November 29, 2005
A housing Bull and Bear
will have different views of
whether spreads are attractive.
-
Valuing Subordinate ABS 3 February 2006
Refer to important disclosures on page 25. 17
When the breakeven spread is above (below) the cash spread, the bond appears
rich (cheap). In this example, the Bull finds all bonds cheap while the Bear finds
all bonds rich.
Impact of Volatility: Thinking of Subordinate Bonds as Options
In the above example, we found that the Housing Bear believes that all of the
bonds are expensive at current levels. Yet, under a 4% HPA scenario (Chart 9)
the Bears mean HPA the Baa2 and even the Baa3 security incur no decline in
price. Why then do they appear so rich according to the Bear? The answer lies not
in the Bears mean of 4%, but in the significant tail of the Bears distribution
(Chart 11, left side). The significant probability that HPA could be lower than 4%
is what makes the Bears distribution indicate that spreads are too tight.
This implies that we need to consider not only the mean of the distribution,
but also its standard deviation. In the prior example, we used a standard deviation
of 7%. What would happen if we reduced this to 4%, thus tightening both
distributions?
The Bull now finds the Baa3 extremely cheap, as it is trading at DM of 300 bp, but
its fair value is 24 bp. Even the Bear, who previously thought that the Baa3
should be trading at 778 bp, would now be content with a spread of just 562 bp
(Table 3).
In fact, with a 4% volatility, the Bear actually believes that the A3 and the Baa1
classes are cheap. Lowering the volatility materially reduces the likelihood of
these classes incurring losses. The Baa1 only begins to lose value at an HPA of
around -4% per year; with the lower volatility this scenario is two standard
deviations below the 4% mean of the Bear (Chart 11).17
Table 3: Housing Bear and Bull Breakeven Spreads (Volatility=4%)
Class MoodysMarket Cash
Price $Market ABS CDS
Spread (bp) Bull B/E Spread
(bp)Bear B/E Spread
(bp)
M6 A3 99-21 80 1 -19
M7 Baa1 99-19 135 -2 2
M8 Baa2 98-04 190 0 208
M9 Baa3 95-25 300 24 562
M10 Ba1 86-02 765 101 1105
Rich Cheap
Source: Merrill Lynch
Having introduced HPA volatility into the analysis, we can now think of
subordinate ABS as being short options on home prices. Remarkably, if we
review Chart 9 and consider the shape of the price distribution for any given class,
we find that it remarkably resembles a put option on HPA (with a few structural
nuances). Consequently, it makes sense that subordinate securities, as with interest
rate options, should be valued using a volatility assumption for home price
scenarios. Each bond in the capital structure corresponds to being short a put
option with a different strike in HPA space. The Ba2 class is struck closest to-the-
money and accordingly offers the greatest spread in return for the greatest risk.
17 As volatility drops, there can even be scenarios in which the Bear would pay more for a bond than
the Bull; indeed, this is shown in the A3 pricing in Chart 9, where the Bear believes the bond is worth
L-19 while the Bull has L+1. The reason is that while both believe that writedowns are highly
unlikely, the expected price of this class is higher at 2% HPA than at 11%. This effect is due to bond
extension arising from lower prepayments and deal paydown rules. In other words, the Bear is
compensated for his HPA brinkmanship he expects HPA to be much lower than it is today but is
confident that it will not go negative by much under this volatility assumption.
Volatility (uncertainty) is an
essential component in risk
valuation.
Holders of subordinate ABS are
writing put options on home
prices.
We begin to consider
subordinate ABS as short
options on HPA.
-
Valuing Subordinate ABS 3 February 2006
18 Refer to important disclosures on page 25.
Market Implied Home Prices
An Analogy With Swaptions: Using CDS to Derive an Implied HPA
Distribution
In the prior section, we imposed our own views of the distribution of HPA
scenarios and calculated fair value spreads, which we then compared with actual
market spreads. However, how are we to know the appropriate distribution of
HPA scenarios? In this section, we consider using the market itself to reveal the
implied distribution.
Readers familiar with interest rate derivatives will immediately recognize this
problem as very similar to the daily calibration of forward curves and implied
volatility to fit observed rate levels and swaption and cap prices. The implied
forward rate curve is not guaranteed to be realized rather it is a convenient
representation of the markets current price for interest rate risk.
In order to apply a similar methodology to home prices, we simply need a family
of spreads referencing different bonds in a single deal, such as the one in our
example; this provides adequate information to deduce a market-implied HPA
distribution. By analogy to the Black model for interest rates, we employ a
normal distribution to model market implied HPA. Using forward interest rates
and volatility from the swaps market and fitting the HPA mean and standard
deviation accordingly, we arrive at an implied distribution of HPA scenarios that
best fits the current ABS spreads.
We label the option-adjusted framework that prices the risk arising from home
prices as well as that arising from interest rates as COAS Credit and (Rate)
Option Adjusted Spread. 18 Both home prices and interest rate expectations are
implied by market prices.
It turns out that the distribution that best fits ABS CDS spreads in our example has
a mean of 8% and standard deviation of 7%, situating it right between our Bull
and Bear investors.19 The three distributions are shown in Chart 12.
Chart 12: Market Implied Home Price Appreciation
0%
1%
2%
3%
4%
5%
6%
7%
-15 -5 5 15 25
HPA (%)
Probability
Implied Forward HPA
Bullish HPA
Bearish HPA
Source: Merrill Lynch
18 Credit used in this context refers to home prices and all correlated drivers of borrower default. Note
also that we allow interest rates to vary as well as home prices. We also allow for correlation between
rates and home prices, with higher rates implying lower rates of growth for home prices, as has been
suggested by empirical evidence. 19 as of November 29, 2005.
The crucial step: infer the
market implied HPA
distribution from ABS CDS
spreads.
The ABS CDS market implied
HPA is centered about 8% per
annum.
-
Valuing Subordinate ABS 3 February 2006
Refer to important disclosures on page 25. 19
This outcome is surprisingly reasonable. ABS CDS spreads are primarily
determined by the balance between synthetic CDO originators sourcing synthetic
collateral and, on the other side, macro hedge funds buying protection thereby
shorting the consumer. The true valuation of these spreads is quite complex: it
requires a model for prepays, defaults, and losses in terms of home prices and
interest rates as well as a model for translating those relationships into price terms.
Yet despite this complexity, the current market equilibrium appears to be at least
roughly in the realm of reasonableness. Of course, aggressive ABS CDS
protection buyers, including some macro hedge funds, might find an implied 8%
( = 6.8%) HPA distribution is far too optimistic. They, accordingly, may want
to purchase protection at todays levels.
Capital Structure Arbitrage
Once the market implied HPA distribution is determined and prices are generated
for each attachment point in the capital structure, comparison to actual prices
provides a natural relative value framework within the capital structure. In
particular, the residuals from the fits used to build the HPA curve provide a rich /
cheap analysis framework. This is because the HPA curve above is the best fit on
average across all the tranches. There remain residual errors on individual
tranches: some appear too tight (rich) while others appear too wide (cheap). We
identify arbitrage opportunities in Chart 13, using pricing (Table 1a) and implied
HPA distribution (Chart 12). The rich/cheap amount in basis points is on the left
hand y-axis and denoted by bars. The overall spread levels are denoted by the line
and presented on the right hand y-axis.
Chart 13: Quantitative Capital Structure Arbitrage
-120
-100
-80
-60
-40
-20
0
20
40
60
80
100
A3 Baa1 Baa2 Baa3 Ba1
Class
Ric
h / C
heap
(b
p)
0
100
200
300
400
500
600
700
800
900
CD
S S
pread
(b
p)
Rich / Cheap
CDS Spread
Source: Merrill Lynch
The Baa2 and especially the Baa3 classes appear rich. This is not a surprise given
the CDO bid for the two classes. At the same time, the Ba1/Baa3 swap appears
170 bp cheap.20,21 One could buy the Ba1 and buy protection on the Baa3 a
positive carry trade, albeit with financing considerations.
Another way to take advantage of the Baa3 richness would be to own Baa2s and
Ba1s versus Baa3s; this also is a positive carry trade which only loses in the
narrow band of scenarios in which the Ba1 incurs losses but the Baa3 remains
mostly whole. 20 We caution that AB CDS spreads were volatile during this analysis and marking an accurate Ba1
spread is difficult due to infrequent trading. 21 Due to funding issues, this cash swap should always be cheap. Typical repo haircuts for the Baa3
class are 15%-25% where as that for the Ba1 are closer to 30%-40%. In theory, one would execute both
legs synthetically by selling protection on the Ba1 and purchasing protection on the Baa3.
Capital structure relative value.
Sell Baa3, buy Ba1 or Baa1.
-
Valuing Subordinate ABS 3 February 2006
20 Refer to important disclosures on page 25.
Housing Risk Metrics
Now that we have presented subordinate bonds as short options and derived an
implied distribution of HPA scenarios from current pricing, it makes sense to
consider the next step. In dealing with interest rate options, we usually consider
their risk in terms of duration, convexity, and vega. Suppose we applied a similar
approach to subordinates, but rather than calculate the exposure to interest rates,
we calculate the exposure to home prices. These metrics applied at the portfolio
level would provide a windfall of valuable exposure information to todays
investor.
We provide an example using our reference deal and the implied HPA distribution
determined in the previous section (Chart 12). The duration (delta) is the
percentage change in a securitys price for a 1% change of the mean of the
distribution, while the gamma represents the convexity associated with moving the
mean by 1% up relative to 1% down (Table 4).
As expected, the ABS are all long delta on HPA they all benefit from a rise in
HPA and all are negatively convex relative to HPA they lose more for a 1%
decline in HPA than they gain from a 1% rise. This is makes sense given that they
are effectively short put options.
Finally, the vega gives the percent change in option-adjusted bond price for a 1%
increase in the standard deviation; as can be seen, the Baa3 has the most vega as
well as the greatest gamma.
It is possible that the spread widening toward the end of 2005 reflected changing
investor views on both the mean and standard deviation of the HPA distribution:
as investors became more bearish on housing and more uncertain about its future,
the price of subordinate ABS declined.
Table 4: HPA Risk Measures
Class Moodys Price ($) Spread (bp) HPA Delta HPA Gamma HPA Vega
M6 A3 99-21 80 0.6% -0.19% -1.3%
M7 Baa1 99-19 135 1.0% -0.28% -1.9%
M8 Baa2 98-04 190 1.9% -0.38% -2.6%
M9 Baa3 95-25 300 3.0% -0.42% -3.0%
M10 Ba1 86-02 765 4.0% -0.36% -2.6%
Source: Merrill Lynch
These risk measures could be used in a number of different ways.
They represent the response of the security to changes in perceptions of the
housing market. In our view, the prices of subordinate bonds should move as
new housing data become available. Of course, changes in HPA are not as
apparent as changes in rates, but on days when home price information is
released, we believe subordinate prices should move. So far, we have not
seen this kind of response in the market, but perhaps that is no surprise as this
methodology is relatively new.
These exposures can be used for constructing trades which are duration-
neutral not only to rates but also to home prices. Eventually, we might want
to calculate similar measures for agency securities as well. For example,
agency IOs have negative HPA deltas a decline in HPA would slow down
speeds and raise the value of the IO. This immediately suggests that
subordinate ABS could be an appropriate hedge for agency IOs; calculating
accurate HPA deltas and gammas would be critical for determining a hedge
ratio.
The Greeks for housing?
-
Valuing Subordinate ABS 3 February 2006
Refer to important disclosures on page 25. 21
5. Final Thoughts and Direction for Future Work
Final Thoughts
Our main findings concerning this new approach to understanding and valuing
subordinate ABS securities are summarized below.
Sub-prime prepayments, defaults, and loss severities are all highly correlated
with home price appreciation. Consequently, cumulative losses for a given
pool are driven in large part by HPA. Using these relationships, we can value
subordinate securities under different HPA assumptions.
Imposing a weighting on HPA and interest rate scenarios generates an option-
adjusted fair value price for each security. This price, in turn, implies a fair
spread on the security for the chosen distribution.
We explore the impact of different distribution means and standard deviations
on valuation, and suggest that investors begin thinking of subordinate ABS as
being short options on home prices.
As a further step, letting the market imply an HPA distribution is similar to
the approach taken with other types of options. An implied distribution
captures the markets pricing of risk and, to some degree, its expectations.
These techniques readily lend themselves to quantitative relative value
recommendations.
This approach naturally sheds light on capital structure arbitrage and relative
value across different tranches. It also allows the calculation of risk exposures
to the housing market.
We hope this discussion and analysis has helped investors gain a better
understanding of ABS subordinates. We plan to work on putting this analytical
framework into automated production in order to track and value a wide variety of
ABS. The recently introduced liquid benchmark ABX index is likely the best
source for calibrating an implied HPA distribution. In addition to these goals, we
believe that there remain a number of possible avenues for further research, and
we now turn to a discussion of these.
Directions for Future Work
There are many possible avenues for further work along these lines. Here we
explore five: i) ABS relative value; ii) cross-sector opportunities; iii) further study
of implied and actual HPA distributions; iv) the prospect of a full Monte Carlo
simulation; and v) exchange traded HPA futures and options.
ABS Relative Value
In this paper, we have developed a new methodology for the valuation of ABS
spreads in a single transaction. By considering collateral, structure, HPA, and
interest rates in an unified framework, we back out an implied home price
distribution given market spreads. As we continue to build our capabilities in this
area, we would like to apply the same technique to a wider variety of structures,
seasoning, and collateral. We suspect that this approach will illuminate some very
interesting relative value opportunities.
A good application is the calculation of fair payups for seasoned bonds, say
between 2004 and 2005 vintage deals. The more seasoned bond would have lower
current LTV and greater subordination, but slightly less structural integrity given
tighter rating agency 2005 rate stresses. Credit-Adjusted OAS would capture the
impact of each in a form that is most appropriate, namely price. It would also
provide a quantitative window on the impact of rating agency guidelines.
-
Valuing Subordinate ABS 3 February 2006
22 Refer to important disclosures on page 25.
Cross Sector Possibilities
Beyond the realm of just ABS, there are possibilities for the extension of this
framework to other areas of the MBS market. Thus far, residential ABS CDS
trading has been restricted to sub-prime transactions. However, home price
information gleaned from spreads using sub-prime models could be useful in
valuating other sectors as well. As an illustration, one could use the implied HPA
distribution extracted from ABS CDS spreads and apply to non-agency
subordinate MBS. In theory, investors could also use the implied HPA
distribution to project a distribution of prepayments on agency pass-throughs or
derivatives. In either of these endeavors, there would certainly be challenges, as
different models could give rise to biases, and the two markets have very different
participants. Nevertheless, adapting and transferring housing market information
embedded in ABS CDS to other sectors may be well worth pursuing.
Further Study of the Implied and Historical HPA Distribution
In calculating the implied HPA distribution from ABS spreads, we made the
simplified assumption of a normal distribution. This could potentially be
incorrect. After all, interest rate traders and researchers still debate over normal or
log-normal rate distributions, even with a quarter century of data and wisdom. We
did test a few other distributions and found few differences in the resulting
valuations.
An even more interesting topic, in our view, is the relationship between the
implied HPA distribution and the actual, historical distribution. Chart 14 shows
the historical distribution of home price appreciation rates for the U.S. since 1980.
The data are recorded quarterly and represent year-over-year HPA.
Chart 14: Historical HPA for the United States (1980 2005)
Source: Merrill Lynch, OFHEO
Comparing the historical distribution to the implied in Chart 12, we find that
todays implied HPA volatility of 7% is more than twice the historical average
over the past 25 years. The implied mean of 8% is also higher that the historical
average of 5.7%. This makes sense: the nation has just enjoyed a sustained period
of high HPA and is at a juncture characterized by uncertainty.
Generally speaking, when implied volatility exceeds actual volatility one
considers selling options, e.g., buying subordinate ABS in our case. However, we
also see that the historical mean of 5.7% is below the implied 8% in Chart 12; this
suggests the opposite trade. The risk measures presented in Table 4 can help
quantify these issues, but we plan to develop this analysis further in future work.
0
5
10
15
20
-6%
-4%
-2%
0%
2%
4%
6%
8%
10%
12%
14%
16%
18%
20%
Year-over-Year HPA
# P
erio
ds
USA
USA Fit Normal
= 5.7%
= 2.8%
-
Valuing Subordinate ABS 3 February 2006
Refer to important disclosures on page 25. 23
It may also be important to consider differences in home price behavior across
geographical regions. As we discussed in earlier, regional differences have
dominated home price behavior across the U.S. The historical standard deviation
of home price appreciation rates in California over the past 25 years has been far
greater than that of the Midwestern states. This may be due to many factors, but
one of them is almost certainly the fact that the supply of land in many areas on
the coasts is limited. Consequently, changes in demand associated with variables
such as income, rates, or consumer confidence will have a greater impact in these
areas than in areas where land is more readily available. We compare year-over-
year historical HPA volatility for Ohio and California in Chart 15.
Chart 15: Historical HPA for Ohio and California (1980 2005)
0
5
10
15
20
25
30
35
40
-6%
-3%
0%
3%
6%
9%
12%
15%
18%
21%
24%
27%
30%
Year-over-Year HPA
# P
erio
ds
OH
CA
OH Fit Normal
CA Fit Normal
Ohio
= 4.5%
= 1.3%
California
= 7.1%
= 8.6%
Source: Merrill Lynch, OFHEO
These types of comparisons and the exposure of the different tranches to both the
mean and the standard deviation could also create some interesting opportunities
for relative value. For example, Ba1 securities are particularly sensitive to the
mean of the distribution because even some low level of positive HPA could bring
these closer to losses. On the other hand, Baa1 securities are more exposed to the
standard deviation, because it is only in the tail of the distribution that they incur
losses, and a lower standard deviation suggests a lower likelihood of a negative
HPA regime. Perhaps one should look for Baa1 and higher-rated securities with
higher concentrations in the low volatility Midwest areas of the country.
Toward a Full Monte Carlo Simulation?
Although we have moved toward a probabilistic HPA approach, we are still using
only static interest rate and HPA scenarios. A stochastic implementation, on the
other hand, would allow home prices and rates to vary over time. It would include
valuable whip-saw scenarios in which both risks vary in time and interact with
bond structures. This would be more analogous to the standard Monte Carlo
simulation approach used for OAS in MBS and CMOs.
Although there are certainly advantages to moving to a Monte Carlo approach,
there are some reasons not to leap to this stage as well:
First, our current approach is more transparent, easier to understand, and
better adapted to the manner in which ABS trade today: static spreads and
speeds. Given that a probability-weighted HPA approach is completely new
to the ABS market, we might not want to go for the full black box approach in
which the results are sometimes more difficult to grasp intuitively.
-
Valuing Subordinate ABS 3 February 2006
24 Refer to important disclosures on page 25.
Second, the calibration of a home price term structure model is likely quite
difficult. After all, unlike interest rate derivatives, existing ABS CDS (or for
that matter existing cash ABS) are not available for an arbitrarily chosen
forward time period. Different models could yield quite different results; for
example, whether HPA is mean reverting or not could become an important
question. Another question is whether one should run HPA series purely on a
national level or on a regional level; if we choose the latter there will be
issues of the correlation between the regions and their individual volatilities.
Nonetheless, a Monte Carlo simulation could be a reasonable next step at some
point in the future.
Potential for Using Traded HPA Futures; The Analogy to MBS
In this paper, we derived an implied HPA distribution from the market prices of
ABS CDS, making an analogy to the common practice of calculating implied
interest rate volatility from the market prices of caps and swaptions. It would be
even more convenient, however, if we could obtain the implied HPA distribution
not from ABS CDS, but rather from an external source that directly references
home prices. When the Chicago Mercantile Exchange begins to trade home price
futures, this may become a real possibility if that market becomes sufficiently
deep and if an options market develops.
If this does become reality, then the treatment of housing risk and interest rate risk
would be nearly symmetric. We would derive both from their respective derivative
markets.
This would greatly enhance the current methodology. Given only one source of
securities which depend on HPA distribution, we are forced to choose between
imposing our own distribution (based on view or history) or using the CDS
spreads themselves to imply a distribution. Neither approach allows us to take an
independent, market implied distribution and assess the Credit-Adjusted Spread of
the ABS based on that distribution. This, in fact, is the standard approach in
valuing MBS, and would be a very exciting development in our view.
The authors would like to thank Tim Isgro for his valuable contributions to
this work.
Analyst Certification
We, Kamal Abdullah, Akiva Dickstein, Shaolin Li, Sarbashis Ghosh and Jonathan
Braus hereby certify that the views each of us has expressed in this research report
accurately reflect each of our respective personal views about the subject
securities and issuers. We also certify that no part of our respective compensation
was, is, or will be, directly or indirectly, related to the specific recommendations
or view expressed in this research report.
-
Valuing Subordinate ABS 3 February 2006
Refer to important disclosures on page 25. 25
Important Disclosures
Merrill Lynch fixed income analysts regularly interact with Merrill Lynch sales and trading desk personnel in connection with their research, including to ascertain pricing and liquidity in the fixed income markets.
Copyright, User Agreement and other general information related to this report: Copyright 2006 Merrill Lynch, Pierce, Fenner & Smith Incorporated. All rights reserved. This research report is prepared for the use of Merrill Lynch clients and may not be redistributed, retransmitted or disclosed, in whole or in part, or in any form or manner, without the express written consent of Merrill Lynch. Merrill Lynch research reports are distributed simultaneously to internal and client websites eligible to receive such research prior to any public dissemination by Merrill Lynch of the research report or information or opinion contained therein. Any unauthorized use or disclosure is prohibited. Receipt and review of this research report constitutes your agreement not to redistribute, retransmit, or disclose to others the contents, opinions, conclusion, or information contained in this report (including any investment recommendations, estimates or price targets) prior to Merrill Lynch's public disclosure of such information. The information herein (other than disclosure information relating to Merrill Lynch and its affiliates) was obtained from various sources and we do not guarantee its accuracy.
Officers of MLPF&S or one of its affiliates (other than research analysts) may have a financial interest in securities of the issuer(s) or in related investments. This research report provides general information only. Neither the information nor any opinion expressed constitutes an offer, or an invitation to make an offer, to buy or
sell any securities or other investment or any options, futures or derivatives related to securities or investments. It is not intended to provide personal investment advice and it does not take into account the specific investment objectives, financial situation and the particular needs of any specific person who may receive this report. Investors should seek financial advice regarding the appropriateness of investing in any securities, other investment or investment strategies discussed or recommended in this report and should understand that statements regarding future prospects may not be realized. Investors should note that income from such securities or other investments, if any, may fluctuate and that price or value such securities and investments may rise or fall. Accordingly, investors may receive back less than originally invested. Past performance is not necessarily a guide to future performance.
The bonds of the company are traded over-the-counter. Retail sales and/or distribution of this report may be made only in states where these securities are exempt from registration or have been qualified for sale. MLPF&S usually makes a market in the bonds of this company.
Foreign currency rates of exchange may adversely affect the value, price or income of any security or related investment mentioned in this report. In addition, investors in securities such as ADRs, whose values are influenced by the currency of the underlying security, effectively assume currency risk.
UK readers: MLPFS or an affiliate is a liquidity provider for the securities discussed in this report. Merrill Lynch Research policies relating to conflicts of interest are described at http://www.ml.com/media/43347.pdf.
Information relating to Non-US affiliates of Merrill Lynch, Pierce, Fenner & Smith Incorporated (MLPF&S): MLPF&S distributes research reports of the following non-US affiliates in the US (short name: legal name): Merrill Lynch (France): Merrill Lynch Capital Markets (France) SAS; Merrill Lynch Dublin (Frankfurt Branch): Merrill Lynch CMB Ltd, Dublin, Frankfurt Branch; Merrill Lynch (South Africa): Merrill Lynch South Africa (Pty) Ltd; Merrill Lynch (Milan): Merrill Lynch Capital Markets Bank Limited; MLPF&S (UK): Merrill Lynch, Pierce, Fenner & Smith Limited; Merrill Lynch (Australia): Merrill Lynch Equities (Australia) Limited; Merrill Lynch (Hong Kong): Merrill Lynch (Asia Pacific) Limited; Merrill Lynch (Singapore): Merrill Lynch (Singapore) Pte Ltd; Merrill Lynch (Canada): Merrill Lynch Canada Inc; Merrill Lynch (Mexico): Merrill Lynch Mexico, SA de CV, Casa de Bolsa; Merrill Lynch (Argentina): Merrill Lynch Argentina SA; Merrill Lynch (Brazil): Banco Merrill Lynch de Investimentos SA; Merrill Lynch (Japan): Merrill Lynch Japan Securities Co, Ltd; Merrill Lynch (Seoul): Merrill Lynch International Incorporated (Seoul Branch); Merrill Lynch (Taiwan): Merrill Lynch Taiwan Limited; DSP Merrill Lynch (India): DSP Merrill Lynch Limited; PT Merrill Lynch (Indonesia): PT Merrill Lynch Indonesia; Merrill Lynch (Israel): Merrill Lynch Israel Limited.
This research report has been prepared and issued by MLPF&S and/or one or more of its non-US affiliates. MLPF&S is the distributor of this research report in the US and accepts full responsibility for research reports of its non-US affiliates distributed in the US. Any US person receiving this research report and wishing to effect any transaction in any security discussed in the report should do so through MLPF&S and not such foreign affiliates.
This research report has been approved for publication in the United Kingdom by Merrill Lynch, Pierce, Fenner & Smith Limited, which is authorized and regulated by the Financial Services Authority; has been considered and distributed in Japan by Merrill Lynch Japan Securities Co, Ltd, a registered securities dealer under the Securities and Exchange Law in Japan; is distributed in Hong Kong by Merrill Lynch (Asia Pacific) Limited, which is regulated by the Hong Kong SFC; and is issued and distributed in Singapore by Merrill Lynch International Bank Limited (Merchant Bank) and Merrill Lynch (Singapore) Pte Ltd (Company Registration No. 198602883D). Merrill Lynch International Bank Limited and Merrill Lynch (Singapore) Pte Ltd. are regulated by the Monetary Authority of Singapore. Merrill Lynch Equities (Australia) Limited, (ABN 65 006 276 795), AFS License 235132, provides this report in Australia. No approval is required for publication or distribution of this report in Brazil.
Highlights1. Introduction2. Today's Housing MarketA Spectacular Run for Home Prices, Especially in Select RegionsChart 1: Home Price Appreciation by State
Is the Housing Market Slowing?Chart 2: NAR Affordability Index: Declining AffordabilityChart 3: Even Seasonally Adjusted, Home Price Growth Has Flattened
3. Home Prices and the Sub-Prime BorrowerChart 4a: Historical Subprime Prepayments by Age and HPAChart 4b: Historical Subprime Defaults by Age and HPAChart 5: Loss Severity versus Home Price AppreciationChart 6: A Schematic of Deal Losses and Home PricesChart 7: Cumulative Loss versus Age by Home Price Appreciation
4. Pricing Housing Market Risk - Credit Adjusted OAS [COAS]Relating HPA Scenarios to ABS SpreadsA Reference DealTable 1a: Ameriquest 2005-R5 Mezzanine StructureTable 1b: Ameriquest 2005-R5 Collateral
Capital Structure Valuation Under Static PricingChart 8: Pricing using HPA, Interest Rates, and a Discount MarginChart 9: Bond Price versus HPA for fixed LIBOR
Toward a Risk-Based Pricing MethodologyChart 10: Option Adjusted Price From Rate and HPA distributionsChart 11: Housing Bear and Bull HPA OutlookTable 2: Housing Bear and Bull Breakeven Spreads [Volatility=7%]Table 3: Housing Bear and Bull Breakeven Spreads [Volatility=4%]
Market Implied Home PricesChart 12: Market Implied Home Price AppreciationChart 13: Quantitative Capital Structure ArbitrageTable 4: HPA Risk Measures
5. Final Thoughts and Direction for Future WorkFinal ThoughtsDirections for Future WorkChart 14: Historical HPA for the United States [1980 - 2005]Chart 15: Historical HPA for Ohio and California [1980 - 2005]
Analyst Certification