Subprime - DuPont Capital Management

advertisement
Subprime:
Sorting Through the Debris
The negative connotation and fear of the ―subprime‖ label often leads to emotionallybased investment decisions, which, in turn, can create opportunity for those willing to provide liquidity. Even in the subprime investment universe, there are no ―bad‖ bonds, if an
investment manager understands the security’s underlying collateral and structural workings. The real issue, however, is whether the bond’s price sufficiently compensates the
holder for the embedded risk. Each security should be evaluated on an individual basis,
because each loan pool has its own nuances, and security performance is often counterintuitive. For example, tranches at the top of the capital stack – with sufficient credit enhancement priced at a discount – could benefit from increased defaults, as long as there is
some recovery value. Recoveries could produce nice gains, as they generate prepayments at
full face value. Tranches at the lower end of the capital structure would fare better if real
estate owned (REO) property processing is delayed, thus allowing these tranches to receive
longer coupon cash flow streams.
We believe, the optimal position in the capital
“ Recoveries could produce
structure (to realize the highest yield) occurs when
nice gains, as they generate
prepayments at full face value.” the investor is repaid his or her last dollar of prin-
May 2010
Volume 2
Karlis R. Ulmanis, CFA
Portfolio Manager
16 Years of investment experience
MBA—University of Southern California
MS—Aerospace Engineering, University
of California
BS—Mechanical Engineering, University
of Illinois
cipal just as the last dollar of credit support is
eroded. However, it is nearly impossible to consistently achieve this because many variables are involved. As is generally the case with investment decisions for a portfolio manager, risk level
should be based on client investment guidelines, and the portfolio’s respective performance
benchmark.
Subprime securities, like many sectors, are typically priced on a grid-like system based on
several factors: weighted-average life, credit support, and collateral characteristics, which
allow traders to quickly bid and offer bonds. However, these factors often fail to address
security-specific traits that can make the assigned price either rich or cheap.
As a portfolio manager, my responsibility is to identify and act on perceived mispricing,
which can help generate excess realized risk-adjusted returns given the embedded downside
-risk exposure. This paper discusses the subprime security characteristics that I, as a portfolio manager, search out (and attempt to avoid!) in my selection process. My goal is to find
value opportunities in the subprime mortgage sector. My current preferences in the subprime market are as follows:
Vintages from 2004 (and earlier)
Sufficient credit enhancement/LTV protection to help absorb existing delinquencies,
and offer a buffer against future defaults
Minimal exposure to secondary liens
Fixed-rate loan collateral
Current pay tranches
Limited geographical exposure to distressed economic areas
Small loan sizes
Subprime securities can be divided into related parts: the underlying pool of loans, or collateral generating cash flow; and the security structure defining how the cash flows and
losses from the loan pool are allocated to the deal’s various investment tranches. I will first
consider the underlying pool of mortgage loans that make up a deal’s collateral, and then
discuss the key characteristics that drive cash-flow dynamics.
DuPont Capital Management
One Righter Parkway, Suite 3200
Wilmington, DE 19803
Tel 302.477.6000
www.dupontcapital.com
Page 1|
The term ―subprime‖ refers to borrowers who have lower
credit worthiness vis-à-vis ―prime‖ borrowers. Credit worthiness is typically measured by Fair Isaac Corporation (FICO)
scores, which range from 300 to 850. The scoring is based
on an individual’s consumer debt record, including credit
card payment history, changes in credit availability, and personal bankruptcy history. Although there is no absolute definition, the subprime range is usually defined by a FICO of
less than 680. Prime loans usually have a FICO greater than
730. A subprime borrower typically pays a higher interest
rate, is required to have a larger down payment, and has more
difficulty refinancing. These disadvantages are magnified,
during times of credit-market stress. Subsequently, subprime
loans behave differently from prime loans, a variable that
must be considered when purchasing subprime mortgage
securities.
Collateral loan pool performance can be characterized by
three parameters: prepayment rate, default rate, and loss severity. The prepayment rate quantifies how quickly principal
flows from the collateral pool, relative to the scheduled amortization rate. The default rate quantifies the rate at which
loans become non-performing and are eventually liquidated.
The loss severity quantifies the realized loss, given that a loan
is liquidated. The expected default rate is a key assumption in
the analysis of a subprime mortgage bond. Two primary drivers of the default rate are the borrower’s ability and willingness to repay a loan. The ability to repay is affected by the
monthly payment amount and economic conditions, such as
income and employment status. A sizable portion of subprime loans originated are based on a hybrid structure (i.e.,
the initial monthly payment was fixed at an artificially low
―teaser‖ rate for two or three years). At the height of the
mortgage lending frenzy, news was widespread about nontraditional 30-year loans called 2/28s or 3/27s. The first
number (i.e., the ―2‖ in a 2/28 loan product) describes the
loan’s fixed-rate period length in years. The second number
describes the adjustable-rate period. Therefore, a 2/28 loan is
a 30-year loan with a two-year fixed rate period, followed by
28 years of adjustable rates.
Another common enhancement to increase the affordability
(i.e., lower the borrower’s monthly payment) was to have a
period of paying interest only. In an interest-only (IO) loan,
the borrower’s entire monthly payment goes toward paying
only the loan’s interest, while maintaining a constant principal
loan balance. Many believe that these ―affordability‖ enhancements allowed borrowers to buy bigger, more expensive homes, which they could otherwise not afford. A downside to these arrangements is that once the teaser feature and
the IO period expired, the monthly payment would often
sharply increase. This occurred as the loan rate became adjustable and principal was added into the monthly payment.
Many subprime borrowers were hit with ―payment shock‖
when their teaser rate expired. Unable to pay their higher
monthly payments, at the height of the housing crisis, many
borrowers defaulted on their loans.
The burden of higher loan payments is a major reason why
the default rate of subprime hybrid adjustable-rate mortgages
(ARMs) is substantially higher than subprime fixed-rate mortgages (FRMs). Exhibit 1 shows a comparison of the constant
default rate (CDR) for outstanding subprime hybrid ARMs
and FRMs. The CDR is computed by annualizing the
monthly default rate. The CDR of outstanding 2003
through2007 (2003-7) vintage subprime FRMs at the end of
2009 was 6.1%, versus 16.7% for subprime hybrid ARMs.
Notably, the CDRs for 2006 and 2007 vintages are substantially higher than for 2003 and 2004 vintages. The 2006 and
2007 vintages of three-year hybrid ARMs (whose teaser fixed
rate periods last expired, or will expire, in 2009 and 2010,
respectively) are now being exposed to payment shock, as
they transition to adjustable rates from artificially-low, fixed
teaser rates.
Exhibit 1. Subprime Default Rate
Subprime Default Rate
20
Hybrid ARM
16
CDR
(%)
CDR(%)
Collateral
The basic understanding of why a borrower repays his or her
mortgage is important for an investor to make a proper assessment of what may influence loan-pool dynamics. Why do
people repay mortgages? There is a combination of reasons:
moral obligation, social judgment, necessity of shelter, accrued home equity, and expectation of future price appreciation. It should be noted that subprime borrowers respond
differently than prime borrowers, a difference that can be
attributed to the former group’s financial limitations.
12
Hybrid
30yr Fixed Rate
8
8
4
4
0
0
2003
2003
2004
2004
2005
2005
So urce: Lo an P erfo rmance, J.P . M o rgan
Source: LoanPerformance, J.P. Morgan
2006
2006
2007
2007
2003-7
Comb
Vintage
Vintage
Economic conditions also affect the borrower’s ability to
repay a mortgage. Reduced income, or a possible job loss,
could force a borrower into default, if they have inadequate
savings to compensate for the near-term shortfall. The geog-
The information contained in this memorandum is intended for the sole use of prospective investors in understanding and evaluating the impact of market events and is not designed or intended to be used for any other purpose. The document may contain forward-looking statements, which are based on current opinions, expectations and projections. We undertake no obligation to update or revise any forward-looking statements. Actual results could differ materially from those anticipated in forward-looking statements. There is no guarantee that any investment in the securities mentioned will be profitable. Investing in sub-prime securities involves risk including the risk of losing some or all of the invested capital. This document is not intended
as an offer or solicitation for the purchase or sale of any security or financial instrument or as a recommendation to invest in any of the securities or financial instruments discussed herein.
Subprime – Sorting Through the Debris
Page 2|
30yr F
raphy of a collateral pool is an important characteristic for
projecting future loan defaults. Exhibit 2 shows the unemployment range by state, as of December 2009. Certain natural disasters (e.g., hurricanes), or industry downturns (e.g., the
automotive industry in Detroit) can create concentrated economic distress. For example, the region comprised of Michigan, Ohio, and Indiana, known as the ―rust belt,‖ experienced elevated levels of unemployment coupled with localizedUnemployment
home price depreciation
(Exhibit 3). Borrowers’ lack of
Rate
income and depleted savings made it difficult to meet
as of December 2009
monthly mortgage obligations. Investors should be concerned about loans in a collateral pool that are concentrated
in economically-depressed regions.
Unemployment
Rate
Exhibit
2. Unemployment
Rate
as of December As
2009
of
December 2009
Las Vegas-based homeowners who purchased their homes in
June 2006 have seen their home values fall 55% through November 2009. If we ignore loan amortization, an original
(June 2006) 80% loan-to-value (LTV) in Las Vegas has morethan-doubled to 178%. The financial incentive to continue
paying on a loan in this situation is essentially nil because the
probability of a full recovery is unlikely in the near term. As
of September 2009, 70% of Las Vegas mortgage holders had
negative equity in their homes, versus 32% for all U.S. mortgage holders. In March 2006, only 5% of U.S. mortgage holders had negative equity.
Earlier vintage loans are more likely to have accumulated
price appreciation and lower current LTVs, which reduce the
likelihood of borrower defaults. This is the primary reason
why I have been focusing on securities with collateral pools
of older (2004 and earlier) vintage loans.
Accumulated
Home Home
Price Appreciation
as of November
2009
Exhibit
3. Accumulated
Price Appreciation
as of Nov’09
Source: Bureau of Labor Statistics
Source: Bureau of Labor Statistics
ource: Bureau of Labor Statistics
Accumulated Home
greater than
Greater
than 10.5%
10.5%
to 10.5%
10.5%
9.0% to
to 9.0%
9.0%
7.5% to
or less
less
7.5% or
greater than 10.5%
9.0% to 10.5%
7.5% to 9.0%
7.5% or less
Home Price
Accumulated
(%)
Price Appreciation
Home Price(%)
Accumulated
Appreciation
Appreciation (%)
40%
40%
Accumulated Home Price Appreciation as of November 2009
20 City Comp
Los Angeles
Miami
20
20City
CityComp
Composite
20%
20%
40%
0%
0%
20%
Detroit
Los
LosAngeles
Angeles
LasMiami
Vegas
Miami
-20%
-20%0%
Detroit
Detroit
Las
LasVegas
Vegas
-40%
-40%
-20%
-60%
-60%
-40%
2000200020012001 2002200220032003 20042004200520052006200620072007 2008
Origination
Origination Date
Date
-60%
2002
2003
2004
2005
Source:2000
Standard2001
and Poors/Case-Shiller
So urce: Standard and P o o rs/Case-Shiller
A borrower’s ability to pay is not always sufficient for a borrower to repay the loan. They must also be willing to pay.
Furthermore, the stigma of defaulting on one’s mortgage has
diminished greatly, as more people now know others – often
family, friends, and neighbors – who have gone into foreclosure, thus making the situation seem less embarrassing . One
of the best indicators of a borrower’s likelihood to be willing
to repay is vested financial interest, or equity in the property.
Equity is the difference between the current value of the
home and the outstanding mortgage loan balance. Equity
increases as the property value increases and/or the loan
balance decreases.
Prior to 2006, the home price appreciation (HPA) over rolling 12-month periods had been always positive. It was easy
for borrowers to stretch for the most expensive home they
could purchase (not necessarily afford) with an expectation
that future price appreciation would remedy the near-term
shortfall. Borrowers who had purchased homes earlier in the
housing cycle had accumulated equity that provided a
―cushion,‖ which they could tap into when faced with financial distress (i.e., job loss, medical bills, etc.). However,
things change. Falling home prices over the past several
years have eliminated, on average, any accumulated price
appreciation, as measured by the 20-city Standard and Poors/
Case-Shiller Composite (Exhibit 3) for homeowners who
acquired their homes since September 2003. For example,
So urce: Standard and P o o rs/Case-Shiller
2006
2007
2008
Origination Date
LTV is a measure of home equity that a borrower has, assuming no other loans on the property. To get a more reasonable assessment of the loan pool loss exposure, the investor needs to adjust the LTV threshold to compensate for the
recovery of servicer advances and property liquidation costs.
In addition to the loan pool’s weighted-average LTV, an investor should look at the upper tail of the LTV distribution
for the loan pool. For example, a pool that is whollycomprised of 50% LTV loans has considerably more loss
protection than a loan pool evenly split between 10% and
90% LTV loans. This is the case even though both pools
have the same weighted average LTV. If all loans in both
pools have a 20% increase in LTV from decreased property
valuations, the default risk increases substantially for the
original 90% LTV loan segment of the pool.
Original LTVs for outstanding subprime loans originated in
the years 2003 through 2007 (Exhibit 4) range from 79-82%
for both ARMs and FRMs. HPA-adjusted LTVs for outstanding loans are substantially higher for 2005 and later vintages because any home price appreciation was overwhelmed
by the subsequent housing downturn. Amortization was not
considered in computing these HPA-adjusted LTVs, which
would only increase the difference between the various vintages, since older loans have had more time to reduce loan
balances.
The information contained in this memorandum is intended for the sole use of prospective investors in understanding and evaluating the impact of market events and is not designed or intended to be used for any other purpose. The document may contain forward-looking statements, which are based on current opinions, expectations and projections. We undertake no obligation to update or revise any forward-looking statements. Actual results could differ materially from those anticipated in forward-looking statements. There is no guarantee that any investment in the securities mentioned will be profitable. Investing in sub-prime securities involves risk including the risk of losing some or all of the invested capital. This document is not intended
as an offer or solicitation for the purchase or sale of any security or financial instrument or as a recommendation to invest in any of the securities or financial instruments discussed herein.
Subprime – Sorting Through the Debris
Page 3|
Exhibit 4. Subprime Loan to Value
Subprime Loan to Value
The characteristics of the underlying loan pool can be key
indicators to how quickly principal will be repaid, as well as
Hybrid ARM
(HPA adjusted) the amount of realized losses. Next, we’ll address how these
30yr Fixed
(HPA adjusted) dynamics may impact the various tranches of securities ofHPA
Adjusted
fered to investment managers.
Hybrid ARM
LTV
(%)
LTV(%)
Hybrid ARM
120
120
30yr Fixed Rate
100
100
30yr Fixed
Unadjusted
80
80
60
60
2003
2003
2004
2004
2005
2005
So urce: Lo an P erfo rmance, J.P . M o rgan
Source: LoanPerformance, J.P. Morgan
2006
2006
2007
2007
2003-7
Comb
Vintage
Vintage
Subprime ARMs were more heavily marketed in the housing
bubble areas, because their ―affordability‖ allowed consumers
to increase access to pricier housing markets and the assumption of future home price appreciation was easier to
make. However, these areas also saw greater home price depreciation when the bubble collapsed. This is confirmed by
the divergence between the ARM and FRM HPA-adjusted
LTV curves in Exhibit 4. Another positive trait of fixed-rate
subprime mortgages is that they are less likely to have a second lien than subprime ARMs. The combined LTV, which
includes the second lien loan amount, is 86.2% for outstanding ARMs originated from 2003 through 2007. This
compares to a combined LTV of only 81.4% for FRMs.
Security Structure
The security structure is layered onto the collateral, or pool of
mortgages. This structure contains the rules for determining
how principal and interest cash flows and losses are allocated
across the various deal tranches. Exhibit 6 provides a simplified view of a typical subprime security. The collateral is depicted on the left side, while the tranched capital structure is
on the right. Higher-quality, less-risky tranches are found at
the top of the capital structure, while lower-rated, riskier
tranches are at the bottom.
Exhibit 6. Typical Subprime Security Structure
Scheduled &
Prepaid (CPR)
Principal
AAA-rated
Sequentials
140
140
Collateral Pool
160
160
Tranche A1
Tranche A2
Tranche A3
Tranche M2
Performing
Assets
Tranche M3
Tranche M4
Exhibit 5. Subprime Loss Severity
Loss Severity (%)
90
LTV < 80%
80
30 days Delinquent
60 days Delinquent
90 days Delinquent
Foreclosure
Real Estate Owned
Liquidation
Liquidations
+ Proceeds
-
Loss
LS
Realized Losses
Severity
Interest
Rate Swap
Tranche M6
Investment Grade
CDR
Constant
Default
Rate
Non-performing
Tranche M5
One relationship that subprime investors should consider –
and one that is less apparent – is that the loss severity of
lower LTV loans is greater than the loss severity for higher
LTV loans (Exhibit 5). This unintuitive relationship arises
from servicers advancing more principal and interest payments for lower LTV loans than higher LTV loans since servicers are less likely to recover their advances for higher LTV
loans. Absent historical loss-severity information for a given
collateral pool, the investor should consider this when making assumptions for loss-severity levels. The availability of
historical data is another positive for focusing on older vintage collateral. Older collateral will have an established track
record, which is directly tied to that collateral and the underwriting standards used to form the loan pool.
Tranche specific
credit enhancement
limit met
Tranche M1
Tranche M7
Trigger
Event
Trigger Events
1) Cum. realized loss
above threshold
2) 60+ day delinquencies
above threshold
Tranche M8
Tranche M9
Tranche M10
Tranche B1
Tranche B2
Overcollateralization
Excess Interest
Stepdown
1) After stepdown date
2) Sr. enhancement% met
3) Above O.C. target
The complexity of a subprime deal keeps many investors
away, particularly those who do not understand the sector, or
have the capability to analyze its structures. As I mentioned
in tmy first paper on the origin of the subprime crisis, I have
seen in the recent past that an ―AAA‖ credit rating is not
sufficient due diligence for placing a subprime security into a
portfolio. While the security’s credit rating can certainly be
used as an identifier of potential problems, it is not the ―all
clear‖ signal we need.
70
60
50
100-120% LTV
40
Source: LoanPerformance, Barclay’s Capital
Aug’09
May’09
Feb’09
Nov’08
Aug’08
30
The security structure can provide investors additional layers
of loss protection, in addition to the LTV embedded in the
underlying loan pool. One protective layer might be a third
party insurance guarantee, which may absorb some or all of
the losses flowing from the loan pool. It should be noted that
this guarantee is only as good as the insurance company offering it. In my experience, I typically do not rely on third
The information contained in this memorandum is intended for the sole use of prospective investors in understanding and evaluating the impact of market events and is not designed or intended to be used for any other purpose. The document may contain forward-looking statements, which are based on current opinions, expectations and projections. We undertake no obligation to update or revise any forward-looking statements. Actual results could differ materially from those anticipated in forward-looking statements. There is no guarantee that any investment in the securities mentioned will be profitable. Investing in sub-prime securities involves risk including the risk of losing some or all of the invested capital. This document is not intended
as an offer or solicitation for the purchase or sale of any security or financial instrument or as a recommendation to invest in any of the securities or financial instruments discussed herein.
Subprime – Sorting Through the Debris
Page 4|
What I have outlined in this paper is a simplified view of my
2,625
1,200
2 yr Auto
5 yr Credit Card
5 yr Manufactured Housing
5 yr Subprime RMBS (right axis)
1,000
800
2,188
2,048
1,750
1,312
600
929
875
400
438
200
Jan’10
May’06
0
79
57
Jul’02
A subprime security structure usually contains additional
complexity, which favors the senior tranches and inhibits
early release of principal to subordinate tranches if the underlying collateral pool is performing poorly. OC that is above
its target level is released to the investment tranches. Prior to
the deal’s stepdown date (usually 3 to 5 years since deal origination), all excess OC will flow to the senior investment
tranches. After the stepdown date – and if the collateral pool
is performing well – a portion of the excess OC will be allowed to flow to the subordinate tranches. Performance criteria typically are based on a check of cumulative pool losses
and a rolling average of the pool delinquency level versus
predefined thresholds or ―triggers.‖ If the underlying loan
pool is performing well (i.e., losses and delinquencies are
below thresholds), the deal is said to be ―passing its triggers.‖
After the stepdown date, the excess OC percentage flowing
to subordinate tranches according to the deal’s schedule increases with time, assuming the triggers continue to pass their
thresholds. This is known as a ―shifting interest‖ structure.
Each deal structure needs to be reviewed on an individual
basis, as any nuances could drastically change the cash flow
dissemination to the various tranches.
Exhibit 7. Asset-Backed Market Spread Histories
Sep’98
I compute an adjusted-credit enhancement level for each
security, which is used as a current state of protection for the
security. To compute this metric, I combine (as a percentage
of the outstanding collateral) the OC and subordination, and
then back out the collateral amount currently 90 days or
more delinquent, the collateral in foreclosure, and real estate
owned. A negative number is troublesome because it indicates an increased likelihood of realized losses, particularly
for longer weighted-average life tranches and pools with high
loss severities.
Because of embedded structural complexities of subprime
securities, (performance triggers, time varying thresholds, and
cash flow/loss allocation rules), I occasionally run across
projected performance which is not intuitive, demonstrating
that unexplored investment decisions can be quite risky. A
collection of analytical tools in addition to investment expertise is necessary to properly evaluate these securities. To formulate and guide our investment decisions, I utilize analytical
systems provided by Intex Solutions, and Bloomberg, L.P., as
well as proprietary investment models.
Nov’94
The third protective layer is credit enhancement through
tranche subordination. The most senior tranche is protected
by other, subordinate tranches. Credit enhancement is typically defined as the quantity sum of subordinate tranche balances and OC, divided by the overall deal collateral balance.
Collateral pool losses will flow to a tranche after third party
insurance has been exhausted, OC has been eliminated, and
all its subordinate tranches have been reduced to zero.
analysis for a subprime security. There are many more issues
that I address in a separate, in-depth analysis. Based on the
collateral pool history, a base case set of assumptions is identified for prepayment, default rate, loss severity, and trigger
projections. This ―base‖ case is not necessarily the most
probable, but more of a severe outcome, which is used to
price the security. Around this base case, several single and
multiple factor sensitivity studies are made, which quantify
how poor the security performance can get.
Spreads to Treasuries (bps)
party guarantees. A second protective layer is overcollateralization (OC). OC is extra collateral in the deal,
which is kept in reserve and is not sold to any investors. OC
can either be placed into the deal at origination, or built-up
over time from the differential between the collateral pool
interest cash flow and the cash flow required to meet interest
payments for the investor tranches. OC is used to offset collateral pool losses before they flow to tranche investors.
0
Source: Deutsche Bank
Current Market Environment
During the past six months, subprime security spreads have
fallen considerably (Exhibit 7). Yields for the subprime security subclass discussed in this paper have fallen to 7% from
12% for 2-3 year weighted-average life issues. Yet, yields for
the sector remain quite attractive, 6% spread over treasuries
and 4% above agency mortgage securities. Currently, I hold
over 30 different subprime securities, which allow odd lot
purchase opportunities at yields between 1% and 2% wider
than market levels. Furthermore, I have completed analysis
of over 50 additional securities, which I would be willing to
purchase at the appropriate yield level. Although risk still
remains and personal circumstances should be carefully considered, I believe that patient investors with access to expertise and analytical resources can still find significant value in
the subprime mortgage sector.
The information contained in this memorandum is intended for the sole use of prospective investors in understanding and evaluating the impact of market events and is not designed or intended to be used for any other purpose. The document may contain forward-looking statements, which are based on current opinions, expectations and projections. We undertake no obligation to update or revise any forward-looking statements. Actual results could differ materially from those anticipated in forward-looking statements. There is no guarantee that any investment in the securities mentioned will be profitable. Investing in sub-prime securities involves risk including the risk of losing some or all of the invested capital. This document is not intended
as an offer or solicitation for the purchase or sale of any security or financial instrument or as a recommendation to invest in any of the securities or financial instruments discussed herein.
Subprime – Sorting Through the Debris
Page 5|
Download