Escolar Documentos
Profissional Documentos
Cultura Documentos
December 2009
PLEASE REFER TO THE LAST PAGE FOR ANALYST CERTIFICATION(S) AND IMPORTANT DISCLOSURES
Barclays Capital | U.S. Securitized Products Outlook 2010
FOREWORD
For over two years now, the US securitized markets have been the center of attention for
much of the world’s investor base, including those who have never bought these securities.
At the start of 2009, the subprime market had already passed into infamy, and there was a
very real risk that other securitized areas would follow. We are happy to report that the
systemic risk posed by securitized markets is now much lower than a year ago.
This has not been achieved without costs. The US Federal Reserve has made herculean
efforts: from spending more than a trillion dollars to buy agency MBS to using the TALF
program to jump-start lending in consumer ABS, the Fed has left no stone unturned. No
doubt, these unprecedented measures have helped avert a systemic collapse, but the
consequences are probably not all that policy makers were hoping for. For example, while
mortgage rates are near all-time lows, mortgage credit has tightened in the past year,
blunting the benefits of record low rates. Losses in non-agency loans are set to be a drag on
bank balance sheets for the next few years, and we expect commercial mortgage losses to
pick up steam from 2010. On the positive side, secondary prices in virtually every securitized
asset are up sharply from their lows, as risk premiums have collapsed, consumer ABS
issuance is a success story, and home prices have been stabilizing for several months
(though we believe there is mild downside still left).
Now that US securitization has pulled back from the brink, where do we go from here? This
publication attempts to answer that question by taking, you, our clients, through our views
on every securitized asset class for 2010. Our analysis is framed against the backdrop of a
recovering economy and a Fed that slowly starts to drain liquidity from the second half of
next year. We discuss issuance estimates, prepayment and default projections, future
losses, and our call for home price moves in various parts of the country. While much of the
easy money in securitized markets has already been made, we highlight areas of
opportunity for next year.
As always, our focus will be the same: to help you make informed investment decisions in
the US securitized markets.
Ajay Rajadhyaksha
Head of US Fixed Income and Securitized Research
Barclays Capital
21 December 2009 1
Barclays Capital | U.S. Securitized Products Outlook 2010
CONTENTS
OVERVIEW
Back from the brink 4
Securitized assets are unlikely to pose systemic risk to financial markets in 2010, unlike
in 2008-09. Most spread products should do well in the first six months¸ but relative
value trades should be the main way to boost returns. A slow draining of liquidity might
be a mild challenge in H2 2010, but upcoming regulatory/legislative changes are bigger
wild cards.
AGENCY MBS
Mortgage basis outlook: Life after Fed 12
Agency MBS will probably adjust well to life after the Fed. The basis should widen around
30 bp, but banks, money managers and foreign investors should then provide a backstop.
RESIDENTIAL CREDIT
2009 a tough act to follow, but upside remains 37
We enter 2010 with an overall favourable outlook. Valuations are still attractive relative to
risky alternatives, and we are seeing the first signs of credit burnout. While modification-
related uncertainty or misdirected government actions pose risks, positive technicals should
overwhelm them, at least in H1 10. As risk premiums compress, deal selection will become
more critical and result in tiering across several dimensions.
CMBS
Mind the gap(s) 61
We begin the year with a positive stance on seniors and select AMs, but negative on
subordinates, as defaults continue to rise given the lagged effect on property cash flows. Toward
H2 10, we see growing pressure as stimulus fades, rates rise, and downgrades mount.
CONSUMER ABS
One year later, the world is a better place 75
Heading into 2010, performance is stabilizing and spreads are tightening. The biggest risks
to the sector in 2010 are regulatory and legislative.
SPECIAL TOPICS
Housing risks and prospects 89
National home prices should fall another 8% in the CS index and trough in Q2 10, thanks to
weak seasonals and the overhang of foreclosed inventory. But continued foreclosure
paralysis has increased the chance of a delayed bottom. We present a delay scenario where
home prices bottom out at similar levels, but in Q2 11.
21 December 2009 2
Barclays Capital | U.S. Securitized Products Outlook 2010
21 December 2009 3
Barclays Capital | U.S. Securitized Products Outlook 2010
OVERVIEW
Monetary tightening, whether in the form of liquidity withdrawal or fed funds rate
hikes (not until the second half), should be gradual and is not expected to disrupt
securitized markets.
Nevertheless, the end of the Fed purchase program should push agency MBS
spreads wider by 30bp.
Our forecast calls for home prices to drop 8% from current levels, before stabilizing
in Q2 10. We expect the macro effect of this decline to be muted.
Commercial mortgages and tight residential credit are set to continue to weigh on
US securitized markets.
But for investors in these assets, there is plenty to talk about. Whether it be a further drop in
home prices, a re-opening of non-agency markets, or discussions about the future of the
GSEs, 2010 promises to be an exciting year. It will just not be the heart-pounding, spine-
chilling excitement that we saw in early 2009; and for that, we should be thankful.
1
We ignore T-bill comparisons between 2009 and 2010 in these numbers.
21 December 2009 4
Barclays Capital | U.S. Securitized Products Outlook 2010
Our rate views assume economic growth of 3.5-4.0% in 2010. We are a little more
optimistic about 2010 growth than the Fed (and some fixed income investors). But given
the depth of this recession, our forecast is mild by historical standards. As a result, we
expect the Fed to start tightening monetary policy only in the second half of the year. The
first step will probably be a slow draining of liquidity through measures like reverse repos.
From September, the Fed should start hiking the funds rate, and take it to 1% by the end of
2010. We then expect the Fed to move to the sidelines in the first half of 2011 before resuming
hikes. In other words, monetary tightening – through the funds rate or other measures –
should be gradual, and is unlikely to cause problems for the securitized markets in 2010.
Agency MBS, and the future of the GSEs: Don’t hold your breath
The end of the Fed purchase One place where Fed changes could cause problems is agency MBS. After all, one of the
program should push agency biggest beneficiaries of Fed intervention has been the agency mortgage basis, with MBS
MBS spreads wider by 25-30bp spreads at their tights for the last few months (Figure 3). The end of the $1.25 trillion Fed
purchase program should pressure spreads wider. But while we expect some widening, we
believe it will be met by strong demand. To understand why, consider the demand-supply
dynamics in 2009. The Fed bought around $1.1trillion in agency MBS in 2009. But net
agency MBS issuance was less than $400bn. That means someone sold the Fed the other
$700bn. One big question is: Who were these sellers and will they come back to buy when
spreads widen? We think the answer is a definite yes.
As Figure 4 shows, most of the selling was from unleveraged accounts such as money
managers and mutual funds. In fact, money managers, mutual funds and insurance
companies together sold over $300bn in the first six months, or an annualized $600bn. And
many of these investors are bench-marked to the Barclays indices. Consequently, most top
fixed income money managers are now sharply underweight MBS against their benchmark
– after all, spreads are so tight that there was little upside in being long. But as spreads
widen with the Fed’s departure, demand should come back as index players move closer to
market weight. Our verdict: spreads should widen 30bp when the Fed walks away before
being met with demand from index money, and the banking system. We discuss details in
Mortgage basis outlook: Life after Fed on page 12.
Figure 1: Net fixed income supply should jump next year Figure 2: Rates forecast for 2010
Fed Funds 3m Libor 2y 5y 10y 30y 10y RY
3,000 2,635 1Q10 0.00-0.25 0.28 1.10 2.50 3.70 4.70 1.40
2,500 2Q10 0.00-0.25 0.27 1.60 3.00 4.20 5.20 1.70
2,000 3Q10 0.50 0.75 2.00 3.40 4.50 5.50 1.90
934 4Q10 1.00 1.20 2.30 3.60 4.50 5.50 1.90
1,500
1,000
500
0
-500
2006 2007 2008 2009E 2009E- 2010E 2010E-
ex ex Fed
Fed/Tsy
Treasury ex-bills
Spread products, term
Net term FI Supply,$bn
21 December 2009 5
Barclays Capital | U.S. Securitized Products Outlook 2010
We do not expect long-term The future of the GSEs is another topic dear to agency MBS investors. By year-end, much of
action on the GSEs for Treasury’s authority to support the GSEs will expire. And the Administration is supposed to
several years present a plan about next steps in February. Sweeping changes are unlikely, but the
government could lay out a timeline for conservatorship and lower the risk weight on GSE
debt. The one place where US Treasury should move quickly is in raising the $200bn each in
preferred backstops. To avoid the tail risk that the $200bn is breached, we recommend
increasing the backstop before year-end (while Treasury does not need permission from
Congress). Indeed, it is possible that the backstop will be raised by the time our readers read
this publication. As for the longer-term future of the GSEs, there are several options: full
nationalization, full privatization, or hybrid models. Each has pros and cons, and we discussed
these in detail in “GSEs: Back to the future,” Securitized Products Weekly, December 11, 2009.
But the history of the GSEs is one of glacial evolution, with Congress often co-opting them for
public policy purposes. This time should be similar, with the legislative focus on health care and
then financial regulatory reform. The Administration will probably preserve the status quo on
the GSEs for a long time, with the final changes emerging many years from now. So for those
waiting for the resolution to the GSE issue, all we can say is: don’t hold your breath.
Home prices, non-agency MBS, and bank losses: Tail risk has declined
Home prices should start falling Unlike with the GSEs, one area where there has been progress is US housing. US home
again, and have another prices, as measured by the Case Schiller index, have stabilized over the past five months. But
8% to go prices could start dropping again soon. While negative seasonals will probably be the
immediate driver (for details, see ‘Housing: Seasonal HPA biases’, Securtized Products
Weekly, July 31 2009), the overhang of foreclosed inventory could play a role (Figure 5). On
the positive side, housing. is being helped by an improving economy, the home-owner tax
credit, and greater affordability. And unexpectedly, home prices have been helped by a new
factor: mortgage modifications.
Modifications have kept To be clear, the modification program (HAMP) is not a silver bullet. As Figure 6 shows,
foreclosed supply from pushing historical re-default rates for all types of modifications are high – HAMP should be better,
down prices but not hugely so. But the process of modification buys time. It increases the number of
months between the borrower turning delinquent and the home hitting the market. This is
shown in the REO (real estate owned) line in Figure 5; even as foreclosures keep rising, the
REO bucket has gone down. So kicking the foreclosure ‘can’ down the road has helped
prices stabilize. Intuitively, if there are millions of foreclosures to still work through the
system, it is better to spread them over a few years than have them hit the market in six
Figure 3: Agency MBS spreads at their tights (bp) Figure 4: Unleveraged investors were net sellers in 2009
EOY 2008 9-Jun Diff
120
Govt. (Fed+ TSY) 71 612 541
100
GSEs (FN, FH & FHLB) 901 1016 115
80 Banks (Commercial,
S&L, & CU) 1030 1122 92
60
Foreign 645 591 -54
40
Mutual Funds 534 400 -134
20 Pension + Retirement Funds 564 494 -70
0 Insurance (Life, P & C) 383 304 -78
Broker Dealers 152 104 -48
-20
REIT 91 103 12
-40 SIV 346 307 -39
Jan 08 Jul 08 Jan 09 Jul 09 Households * 201 107 -94
FN CC Libor OAS Total** 4917 5160 243
Source: Barclays Capital Source:: Flow of funds, FDIC, FN/FH Monthly Summaries, TIC, Morningstar
21 December 2009 6
Barclays Capital | U.S. Securitized Products Outlook 2010
Figure 5: Foreclosures have not led to housing supply Figure 6: HAMP is not a silver bullet
1,000 20%
500 0%
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
-
Jan-05 Jan-06 Jan-07 Jan-08 Jan-09 Months since modification
months – this prevents prices from over-correcting to the downside. And with the
Administration focused on modifications, we expect long delinquency-to-liquidation
timelines to help home prices. As a result, our forecast calls for prices to drop 8% from
current levels, before stabilizing in Q2 2010. The macro impact of this decline should be
muted. After all, a house worth $100 is now worth $67 (prices have fallen around 33% from
peak in Case Schiller). A further 8% decline from current levels is simply another $5.3. As
every month passes without a sharp increase in the REO bucket or a sharp drop in home
prices, the tail risk posed by housing declines ever so further.
Tail risk is also reducing in non-agency MBS. For the first time in a long while, there are
signs of credit burn-out. Current to delinquent roll rates have stabilized in recent months in
sub-prime, partly owing to burnout. There is some good news on severities as well; some
servicers are turning to short sales to avoid liquidation costs and recent severity data show a
down-trend in most agency sectors. Admittedly, while tail risk is receding over time, our
base case loss estimates are not much better than earlier in 2009. There will be more losses
in non-agency MBS – and lots of them.
The banking system still has And these losses will impact US banks. The US banking system has over $13 trillion in balance
significant losses to absorb, but sheet. But future losses will come mainly from the $7 trillion+ loan book. Figure 7 is our forecast
not big enough to be systemic for future losses by loan type; it shows that banks have another $850bn+ in losses. These
21 December 2009 7
Barclays Capital | U.S. Securitized Products Outlook 2010
should be spread over the next few years, at $250-300bn in charge-offs per year. Any such loss
estimate involves many assumptions. But look back at 2009, and our numbers seem to fit. The
banking sector has taken $230bn in losses in Q1-Q3 2009, or about $300bn annualized, very
much the pace we forecast months ago in ‘Overview: Bad Bank Blues,’ Market Strategy
Americas, January 30, 2009. So is there systemic risk to US banks due to future losses? We think
not, assuming 2009 is any indication. Banks have taken $300bn annualized in charge-offs this
year, raised another 75bn annualized in loss provisions and still reported a small profit.
Assuming the funds-10s curve stays steep for another 18 months (we expect it will), banks
should earn their way out of this hole. And the hole is not $850bn, since banks now have
$220bn in loss provisions built up. So if the banking sector can earn $600bn+ over the next few
years, they should be able to absorb losses without eating into the capital base. In addition, we
expect considerable dispersion in losses across the banking system, with regional banks hit
harder than big money center banks (where systemic risk truly resides). Hence, while losses
might pose challenges to the equity side, they do not appear big enough to be systemic.
Residential mortgage credit Another problem area, for both MBS and housing, is the tightening in residential mortgage
should stay tight for the next credit. Figure 9 shows that the average FICO score for loans guaranteed by Fannie Mae and
several months, before starting Freddie Mac has risen sharply in the last year. Agency prepayment data paint the same
to ease picture – the ability to voluntarily refinance (even for good credit, lower LTV borrowers) is
now lower than in the past several years. Why is this happening? Definitive answers are hard
Figure 8: CRE should lag the economic recovery Figure 9: Mortgage credit has tightened sharply
Note: Shaded areas are recessionary periods. Source: PPR, BLS Source: Fannie Mae, Freddie Mac, Barclays Capital
21 December 2009 8
Barclays Capital | U.S. Securitized Products Outlook 2010
to come by. But as we discuss in “Agency underwriting: When will the squeeze ease?” on
page 97’, it might be due to skittish lenders hit hard by loan repurchases, combined with the
need for banks to revamp underwriting platforms. Our research suggests that agency MBS
credit could remain tight for several months, especially since FHA is starting to tighten
standards. But by the second half of next year, standards could start to ease slowly, pushed
along by Administration pressure and banks improving their origination platforms. This has
valuation implications, and not just for agency MBS. For example, even a few CPR of added
prepayments could have a meaningful effect on some non-agency sectors as seen in ‘Sub-
prime speeds: Assessing credit curing“ on page 107.
…especially in H1 2010, before Within spread products, there should still be one common theme. Governments all over the
the Fed starts draining liquidity world are issuing record quantities of sovereign debt. By comparison, spread product
issuance will be low. This should support spread products, especially in the first half of the
year. The agency MBS basis is an exception and we expect it to widen, but that is because of
current valuations. While there are several ways to short the basis, we like being long IOs
hedged with Treasuries owing to the attractive carry. CMM-CMS FRAs are another option;
the carry is not as attractive as in the IO trade, but it is a clean way to position for wider
spreads. We also like taking IO risk in the coupon stack and suggest being overweight the
5.5% and higher coupons. And we expect GN premiums to underperform FN premiums.
Meanwhile, in the non-agency markets, the easy money has already been made. But strong
technicals, such as PPIP buying and the return of third party repo, should spur further yield
compression. We recommend an outright long in alt-A Hybrid SSNRs and also a long in
Jumbo Fixed SSNRs but with repo leverage. Investors positioning for recovery trades should
look to subprime LCF, option ARM SSNRs, and Alt-A Hybrid Mezzanine AAAs. The
tightening trend should stay intact in consumer ABS as well, with mezzanine and
subordinate credit card ABS offering the best value. Finally, we prefer relative value trades in
CMBS, such as buying AJ.2 versus AJ.5 and A.4 versus A.3. We also like being outright short
the A.3 series.
Most of our trades should play out over the next few months. The second half might pose
more challenges, as the Fed starts draining liquidity. But we expect any such move to be
slow. Spread products should be helped by plentiful liquidity for the next several quarters.
21 December 2009 9
Barclays Capital | U.S. Securitized Products Outlook 2010
GSEs plan to consolidate their guarantee books in 2010. But this should not materially
impact agency MBS issuance, because GSE regulatory capital requirements have already
been waived. Away from the GSEs, many other securitized assets might get affected. The
capital relief and the limited liability nature of securitizations could well disappear as off
balance sheet assets move on balance sheet. And the need to hold capital against these
assets under new capital regimes (such as Basel II) could make new issuance difficult in
areas such as credit cards, non-agency mortgages, etc. But it is difficult to say anything
definitive yet, with the language still in flux. And accounting interpretations will play a big
role. For example, it might seem that banks will have to consolidate all RMBS securitizations.
But this can be accommodated by aggregating a deal using loans from several originators. If
no one services a majority of the deal (since there are several servicers), and if banks can
keep a 5% ‘representative sample’ instead of a ‘vertical slice’ of the deal, no firm might have
to consolidate. As a result, the consolidation-related picture should clear only some months
down the line.
Meanwhile, one risk to our base case is credit risk. Sovereign credit should be an important
topic in 2010. The recent example of Greece, as well as the widening of CDS in several
developed countries, shows this issue coming to the forefront. Ironically, with the
developing world a net exporter of capital to the developed world, sovereign risks now seem
concentrated in some big developed economies. How will this impact US assets? We believe
that increased fears about G8 sovereign risk will help US Treasuries, which could widen
spreads. Clearly, we do not anticipate serious concerns about US credit risk in 2010. Our
analysis shows that a big driver of ratings tends to be the share of a country’s currency in
total world reserves. By that metric, the US has a long way to go before other factors (such
as financial stabilization cost, debt/GDP ratios, etc.) drag down the US credit rating (for
details, see “Safe for now,” Market Strategy Americas, July 17, 2009).
The big-is-better argument also holds true for the banking system. Just as the US rating
benefits from the size of the US economy, so do big banks. Consequently, the rating agencies
have started making noises about clauses in H.R.4173, which aim to minimize the impact of
failing financial institutions on the taxpayer. S&P put out a note on December 16, 2009
warning that the bill in its current form might make them take a second look at the existing
ratings of big banks. In other words, if banks are no longer too big to fail, their credit rating
could be negatively impacted. We expect several iterations and discussions between rating
agencies and lawmakers before any language becomes law. But this does increase the risk of
downgrades to parts of the banking system, and so investors should monitor this closely.
Finally, watch out for new policy changes from Washington on the mortgage front. If HAMP
does not work well (as we expect), and foreclosures keep rising, Congress might revisit
some of the more radical suggestions from earlier this year, such as cram-downs, forced
21 December 2009 10
Barclays Capital | U.S. Securitized Products Outlook 2010
debt forgiveness, etc. On the agency MBS side, one tail risk is the prospect of an off-market,
low mortgage rate provided by the government. MBS investors fearful of this shift
compressed the coupon stack sharply in Q1 09 – if such an off-market rate is actually offered
by the government, it could greatly hurt premiums and, thus, all agency MBS valuations.
Another risk is that mortgage credit does not loosen at all for all of 2010, though we think that
is unlikely. In general, we believe that a recovering economy should provide cushion against
radical policy changes, in both agency and non-agency MBS.
All in all, we do not see the risks to our base case to be far greater than in any “normal”
year. And the risks are certainly lower than at the end of 2008, when the only constant
seemed to be rapid change.
21 December 2009 11
Barclays Capital | U.S. Securitized Products Outlook 2010
AGENCY MBS
We expect spreads to widen 30bp from current levels when the Fed purchases stop
Nicholas Strand but then to be met by strong demand. Fed purchases have had other positive effects,
+1 212 412 2057 such as richer rolls and lower mortgage volatility. In the absence of the Fed backstop
nicholas.strand@barcap.com and related benefits, agency MBS look rich. This is despite low supply - new issuance
should stay muted at $350-400bn for 2010, but pay-downs from Fed and GSE
Matthew Seltzer
portfolios should also add to supply.
+1 212 412 1537
matthew.seltzer@barcap.com We think that banks, money managers, and foreign investors should all be buyers of
MBS in 2010, albeit at wider spreads. A host of factors – muted loan demand, low
Philip Ling appetite for credit risk, a steep yield curve, large cash holdings, and shortage of other
+1 212 412 3202 spread products – point to increased MBS demand from banks. International demand
philip.ling@barcap.com could also rise as risk aversion abates and reserve growth picks up.
Despite strong demand, MBS look rich to Treasuries, agency debt, and corporates
on both an absolute and a historical basis. We believe active money manager
participation will be needed to backstop spreads. Our analysis suggests that
money managers should provide a backstop when spreads are 30bp wider, which
is where we expect them to stabilize.
The risks and returns for the MBS markets seem well balanced in 2H10. The
enormous Treasury supply and the shortage of spread assets should help the basis.
But potential liquidity withdrawal could increase volatility and drag down
performance.
Figure 1: Fed purchases of agency MBS ($bn) Figure 2: Net issuance ($bn)
35 1,200
$bn 600
30 1,000
500
25 41
800 178
400
20 176
600
15 300
135 213
400
10 200
200 272 91
5 100 206
0 0 80
0
Jan-09 Apr-09 Jul-09 Oct-09
2007 2008 2009 YTD
Weekly Net Purchase Cum Net Purchases, RHS FN FH GN
Source: Federal Reserve, Barclays Capital Source: Fannie Mae, Freddie Mac, Ginnie Mae, Barclays Capital
21 December 2009 12
Barclays Capital | U.S. Securitized Products Outlook 2010
Figure 3: Mortgage current coupon OAS (bp) Figure 4: Sector excess performance versus Treasuries (bp)
-50
Dec-99 Dec-01 Dec-03 Dec-05 Dec-07
Libor OAS TSY OAS
Source: Barclays Capital Note: 2009 values are through November 2009. Source: Barclays Capital
While the Fed’s emergence as a buyer of mortgages was meant to drive rates lower and
foster new purchase activity, net issuance of agency MBS was actually lower than in 2007-
2008 (Figure 2). Many competing forces, such as tighter underwriting standards and a
weak housing market, contributed to lower net issuance. We direct readers to “Agency
underwriting: When will the squeeze ease?” on page 97 for a more complete discussion.
2
Please see The day the Fed stood still, October 9, 2009.
21 December 2009 13
Barclays Capital | U.S. Securitized Products Outlook 2010
Figure 5: Conventional 30y float stands at only $1.6trn Figure 6: Implied financing rates have dropped sharply (%)
1000 2.50
2.00
800 1.50
1.00
600
0.50
400 0.00
-0.50
200 -1.00
-1.50
0 Nov-08 Feb-09 May-09 Aug-09 Nov-09
<4 4 4.5 5 5.5 6 6.5 7 >7
1m Libor Implied Roll Funding
Float CMO Fed
Source: Federal Reserve, Fannie Mae, Freddie Mac, Barclays Capital Note: Implied funding calculated on FNCL 5.0s. Source: Barclays Capital
Figure 7: Mortgages have traded to short empirical durations Figure 8: Limited carry without Fed support (32s/month)
21 December 2009 14
Barclays Capital | U.S. Securitized Products Outlook 2010
In the second half of the year, the Fed will probably begin to drain liquidity from the system.
We believe reverse repos are the likely tool of choice. Meanwhile, direct asset sales from the
Fed’s portfolio seem extremely unlikely, at least in 2010. Given the amount of MBS the Fed
owns, selling bonds outright could pressure the basis considerably and de-value the rest of
its agency MBS holdings.
The GSEs face a mandate to reduce their retained portfolios by 10% per year. They will likely
find it difficult to sell their non-agency MBS or whole loan holdings, which may put more
pressure on their agency MBS portfolios (Figure 11). Overall, we expect the GSEs to
contribute an additional $10-12bn of supply per month, or $150bn over the course of 2010.
Admittedly, the 10% rule can be changed by Treasury without congressional approval
which could ease GSE selling pressure, but this is not our base case expectation. On the
other hand, GSE buyouts of delinquent loans could pick up in early 2010 (Figure 12). In this
case, the need to make room for delinquent loans in their retained portfolio could cause the
GSEs to sell MBS, which could exacerbate basis widening and volatility.
Figure 9: Agency net monthly supply over the past two Figure 10: Fed portfolio coupon distribution
years ($bn)
80 350
60 300
250
40
200
20
150
0
100
-20
50
-40 0
Jan-08 Jul-08 Jan-09 Jul-09 3.5 4 4.5 5 5.5 6 6.5
FHLMC FNMA GNMA Fed Purchases ($bn)
Source: Fannie Mae, Freddie Mac, Ginnie Mae, Barclays Capital Source: Federal Reserve
21 December 2009 15
Barclays Capital | U.S. Securitized Products Outlook 2010
Figure 11: Composition of GSE retained portfolio ($bn) Figure 12: GSE delinquency pipeline (%)
600
4%
3%
400
2%
200
1%
0 0%
FN FH Jan-06 Apr-07 Jul-08 Oct-09
Agency MBS Non-agency MBS Whole loans FNMA FHLMC
Note: As of October 2009. Source: FN/FH Monthly Summary Source: FN/FH Monthly Summary
Finally, we expect $350bn in net issuance, mainly from home sales. Add the amounts coming
from the GSEs and the Fed, and we end up with roughly $625bn (350 + 125 + 150) for the
market to absorb next year. What are the risks to this forecast? First, changes in underwriting
standards (as discussed in “Agency underwriting: When will the squeeze ease?” page 97)
could affect issuance. For example, tighter FHA standards could limit GNMA supply, while
lower standards from the GSEs could shift more production into FN/FR securities. Second,
higher rates could reduce affordability, pushing supply projections downward.
A) Risk appetite and loan demand: During periods of heightened risk aversion, banks
generally prefer to hold securities with limited credit risk. Underwriting tends to tighten,
causing loan demand to drop off.
B) Attractiveness of securities versus loans: When the yield curve is steep, banks move
away from loans to securities, as they can meet their NIM targets without taking credit
risk (Figure 14). And when the front end is low, current yields on C&I loans are limited,
since they are mostly floating-rate loans.
These factors have meant that over the past five recessions, the banking system increased
its security holdings by around 15% in the year following the recession (Figure 13). On a
balance sheet level, banks have typically increased security holdings by 2-3% of total
balance sheet in the first year after a recession. With their current security holdings close to
$2trn and a balance sheet of around $13.5trn, we estimate that banks could add $300-
400bn in securities during 2010.
21 December 2009 16
Barclays Capital | U.S. Securitized Products Outlook 2010
Figure 13: Bank securities versus loans, 1y change (%) Figure 14: Bank MBS holding versus curve
25% 3 60,000
15% 2 40,000
5% 1 20,000
0% 0.5 10,000
-5% 0 -
Source: Federal Reserve H8 Data, Barclays Capital Source: FDIC, Barclays Capital
Shortage of other spread What percentage of this security buying will be in agency MBS? Historically, banks have held
products argues for a higher close to 50% of securities in agency MBS pass-throughs and CMOs, equating to $150-
allocation to agency MBS 200bn in bank demand for MBS in 2010 (Figure 15). The shortage of other spread products
(non-agencies, CMBS, ABS, etc.), however, may argue for a higher allocation to agency MBS
during this cycle. Also, the banking system is currently flush with cash (Figure 16). This
should also encourage the move into agency MBS. We believe that a good portion of this
buying will be in short-duration assets (hybrids, short CMOs, and 15y), given concerns
about extension risk.
Figure 15: MBS+CMO % of securities, bank portfolios Figure 16: Cash % of total assets, bank portfolios
60% 12%
50% 10%
40% 8%
30% 6%
20% 4%
2%
10%
0%
0%
1992-05 1995-11 1999-05 2002-11 2006-05 2009-11
Q1/1994
Q3/1996
Q1/1999
Q3/2001
Q1/2004
Q3/2006
Q1/2009
Cash as % of assets
21 December 2009 17
Barclays Capital | U.S. Securitized Products Outlook 2010
securities. In addition, $1.2trn of short-term securities are expected to roll off during the
year. If these two cash flows are reallocated in line with current holdings, foreign investors
could add around $100bn in agency MBS in 2010.
Figure 17: Foreign holdings of MBS have stabilized, but look to pick up in 2010 ($bn)
Type of security Jun-05 Jun-06 Jun-07 Jun-08 Dec-08 Mar-09 Oct-09
Figure 18: Supply vs demand, post-Fed purchases (monthly) Figure 19: Money manager MBS allocation
Foreign $10Bn Western Asset Core Plus Bond $7.7 40% 60% -20%
Openheimer Strategic Income $8.1 9% 12% -3%
Money Managers ???
Fund
Lord Abbett Bond Debenture $7.1 2% 14% -12%
Source: Barclays Capital Note: We use the latest holdings available for the largest eight funds indexed to
the Barclays US Aggregate Index. Source: Fund Quarterly Statements, Barclays
Capital
21 December 2009 18
Barclays Capital | U.S. Securitized Products Outlook 2010
Figure 20: MBS spread to agency debt (agency OAS, bp) Figure 21: MBS and high quality credit spreads (bp)
80 400
40 300
200
0
100
-40
0
-80
-100
Oct-99 Oct-01 Oct-03 Oct-05 Oct-07 Oct-09
-120
Nov-02 Nov-04 Nov-06 Nov-08 CC OAS (bp) Credit Spread(bp)
MBS looks rich to Treasuries, When we compare agency MBS with agency debt, the spread (adjusted for optionality) has
corporates and agency debt, historically been close to zero (Figure 20), but it has tightened since the start of the Fed
both on an absolute and purchases. But even though we have seen some moderation, MBS still look 40bp tight to
historical basis agency debt. Mortgage Treasury option-adjusted spreads also look significantly rich
compared with historical levels (Figure 3). We also compare MBS with corporate debt and
look at the spread to Treasuries for the top 20% industrial names (to limit the effect of
credit risk) (Figure 21). These names have historically traded 30bp back of agency MBS,
compensating for lower liquidity. Currently, the spread is close to 100bp, although some of
this due to the credit cycle. In summary, we believe that MBS spreads need to widen 30-
40bp before money managers add MBS in size.
We do not expect spreads to “blow out” much beyond 30-40bp for a number of reasons:
1) money managers are currently underweight and should add at wider spreads; 2) the Fed
could step in if mortgage rates rise too fast; 3) the enormous liquidity in the system should
keep spread assets well bid; and 4) the GSEs could provide a local backstop if MBS cheapen
significantly.
21 December 2009 19
Barclays Capital | U.S. Securitized Products Outlook 2010
On the other hand, there is technical risk to our short basis call. There is a lot of short
interest, both implicit and explicit, in the agency basis now. Much of this has come from
crossover investors in equities and other parts of fixed income. A sharp rally could force
these shorts to cover, further richening MBS. Also, although it is unlikely, in our view, there
is an outside chance that the Fed program could be extended. Both are risks to keep in
mind, but we do not expect either to play out.
For H2 10, we think that On the GSE front, we believe that FAS 166/167 should have limited effect on buyouts (for
risks and rewards are details, please see “Less than meets the eye,” Securitized Products Weekly, November 20,
well balanced 2009 But if the GSEs decide to buy out delinquent loans aggressively, the coupon stack will
obviously compress quite a bit. Another factor to keep in mind is the cheapening of rolls. We
see the removal of excess reserves by the Fed (in 2H10) as the primary driver of roll
cheapening, not the end of Fed purchases, so this is not a story for the next few months.
In fact, most of this article has focused on H1 10. Looking into H2 10, we think that risks
and rewards are well balanced. On the one hand, the enormous Treasury supply and the
shortage of spread assets should cause risky assets to outperform. On the other hand, if the
Fed starts to remove liquidity, then volatility could pick up and mortgages could
underperform. Either way, it promises to be an exciting year for agency MBS.
21 December 2009 20
Barclays Capital | U.S. Securitized Products Outlook 2010
AGENCY MBS
Nicholas Strand GSE buyouts should significantly steepen the conventional S-curve and boost the
+1 (212) 412 2057 speeds of credit-impaired collateral. But an extreme, binary buyout scenario is very
nicholas.strand@barcap.com unlikely.
Conventional discount speeds should trend significantly below existing home sales,
reaching only 4-5 CPR. If mortgage rates back up substantially, the average life of
the 2009 production FNCL 4.5s should extend to about 11.5 years.
In contrast, high involuntary prepays in GNMA pools should give them extension
protection. Even in a 300bp sell-off, the 2009 GNMA 4.5s should not extend beyond
7.5 years.
Figure 1: 2009 prepays were far below historical levels Figure 2: S-curves of loans with current LTV>80%
CPR
CPR of FNMA 30y universe
50
70 2000-2008
60 2009 40
10 0
0 0 25 50 75 100 125 150
-100 -75 -50 -25 0 25 50 75 100 125
FN 30y Incentive to no-point mortgage rate FICO < 720 720-770 > 770
Source: FNMA, FHLMC, Barclays Capital Note: 9-30 WALA FHLMC loans. Source: FHLMC, Barclays Capital
21 December 2009 21
Barclays Capital | U.S. Securitized Products Outlook 2010
Figures 2 and 3 show S-curves during 2009 for different LTV-FICO stratifications. Even for
loans with less than 80% LTV, the speed differential between 770 and 720 FICO has been
abnormally large. Historically, the median FICO of the US population has been 723. But
extremely tightened underwriting during 2009 has pushed the average origination FICO to
760. As lenders re-calibrate their underwriting standard and vie for more volume, we expect
a moderate downward migration in FICO starting in H2 10.
Origination FICO could Consequently, low LTV, mid-tier FICO (that is, 700-740) loans could start exhibiting a
start normalizing toward greater refinancing response in the second half of 2010. High LTV loans, on the other hand,
the end of next year should take longer for refinance-ability to improve meaningfully. Although the HARP
program is in effect until mid-2010, so far it has done almost nothing to help high-LTV
borrowers. As the government seems to have given up on HARP to focus on HAMP, we
expect a negligible effect of HARP on prepayments during 2010.
Figure 3: Opposing forces: Prepayments in 2010 Figure 4: Timeline of delinquency buyouts (FNMA 2006 6s)
30 8
6
20
4
10
2
0
0 25 50 75 100 125 150 0
FICO < 720 720-770 > 770 Sep-09 Jun-10 Mar-11 Dec-11 Sep-12 Jun-13 Mar-14
Note: 9-30 WALA FHLMC loans. Source: FHLMC, Barclays Capital Source: Barclays Capital
21 December 2009 22
Barclays Capital | U.S. Securitized Products Outlook 2010
the pipeline with a serious delinquency rate over 6% (by balance). This trend will almost
certainly change in 2010, as the ramp-up of the HAMP program triggers multiple forms of
delinquency buyouts.
Buyouts should peak in H1 10, Although FNMA has the option to repurchase a loan as soon as it becomes four months
then drop a bit before going up delinquent, its current practice is to buy out a loan only when it is modified or there is a
steadily over the next two years home forfeiture action (short sale, deed-in-lieu, or foreclosure sale). As discussed
previously,3 we expect 20-25% of the existing delinquency pipeline to come out in the form
of a successful modification and the rest to be via home forfeitures. Because modifications
take much less time to complete than foreclosures, buyouts should first spike and then abate
as some of the existing delinquencies are successfully modified. Afterwards, home forfeitures
should put buyouts on a steady upward ramp. By overlaying the modification and foreclosure
components, we arrive at the total buyout timeline (Figure 4): it should peak in H1 10, then
drop a bit before going up steadily over the next two years.
Figure 5 shows our estimated serious delinquency rates by balance for various 30y and 15y
cohorts. Unsurprisingly, defaults are most concentrated in newer vintages and higher
coupons (6s and above). For example, we estimate that 13.6-18.5% of the 2006-08 FNMA
6.5s are seriously delinquent. For the 6s, the numbers are 9.2-10.8%. Because involuntary
prepays should be proportional to delinquency rates, this implies a sharp steepening in the
S-curve.
Figure 5: Estimated serious delinquency rates for FNMA/FHLMC 30y and 15y cohorts
30y fixed 15y fixed
Vintage Coupon FHLMC FNMA Coupon FHLMC FNMA
2003 5.0 0.9 1.0 4.5 0.2 0.2
5.5 1.3 1.6 5.0 0.3 0.4
6.0 2.2 2.7 5.5 0.5 0.6
2004 5.0 1.6 1.9 4.5 0.3 0.4
5.5 2.4 2.9 5.0 0.5 0.7
6.0 3.5 4.2 5.5 0.8 0.9
2005 5.0 3.2 4.0 4.5 0.7 0.8
5.5 4.9 6.1 5.0 1.1 1.4
6.0 7.6 9.4 5.5 1.7 2.1
2006 5.0 3.9 4.9 4.5 0.9 1.1
5.5 5.8 7.2 5.0 1.4 1.8
6.0 7.4 9.2 5.5 1.7 2.2
6.5 11.0 13.6 6.0 2.6 3.2
7.0 16.2 20.1 6.5 3.0 3.8
2007 5.0 3.9 4.9 4.5 0.8 1.0
5.5 5.6 7.0 5.0 1.4 1.7
6.0 8.7 10.8 5.5 1.6 2.0
6.5 14.9 18.5 6.0 2.8 3.5
7.0 23.8 26.3 6.5 5.9 7.3
2008 5.0 1.2 1.5 4.5 0.3 0.3
5.5 2.8 3.4 5.0 0.4 0.5
6.0 4.8 6.0 5.5 0.7 0.8
6.5 9.3 11.5 6.0 1.2 1.5
7.0 15.8 18.2 6.5 4.7 5.8
Note: estimated delinquency rates by outstanding balance as of October 2009. Source: Barclays Capital.
3
See Prepayment Outlook: Delinquency Buyouts: A Brave New World, July 24, 2009
21 December 2009 23
Barclays Capital | U.S. Securitized Products Outlook 2010
While some investors are concerned about the possibility of massive GSE buyouts due to
the rollout of FAS 166/167, we believe that is an unlikely event, for several reasons:
Implementation of FAS 166/167 merely increases the maximum possible savings for the
GSEs from $5bn to $10bn per year.
Large-scale buyouts would make it difficult to comply with GSE portfolio limits. The
GSEs are required to reduce the size of their retained portfolio by 10% on an annual
basis. Delinquent loan buyouts would count toward this portfolio cap.
Funding $250bn in buyouts is not trivial. This would require either a huge increase in
agency debt issuance or the GSEs to sell a substantial amount of MBS.
Both of these funding options could lead to market disruptions, precisely at the time
when the Fed is ending its agency debt and MBS purchase programs.
It makes sense to leave loans that are headed for foreclosure in the MBS trust. Since the
loan is extinguished after foreclosure sale and no longer counted toward the portfolio
cap, why bother buying it out? If it is going to disappear in a year anyway, buying it out
now may not save enough to justify the effect on the financial statement; the portfolio
cap; and the costs of funding, hedging, and managing these delinquent loans.
A sudden, massive buyout In the final scheme of things, a $5-10bn savings per year is overwhelmed by other factors.
would significantly disrupt the The average dollar price of the agency MBS universe is near $105, and a sudden, massive
MBS and housing markets buyout would significantly disrupt this market, the only mortgage market that is still
functional. In addition, the GSEs own about $900bn agency MBS themselves. If massive
buyouts hurt average dollar price by 1 point, it would lead to a $9bn loss to the GSEs.
2007 FNMA 6.5s and 7s Figure 6 shows our expected 1y involuntary speeds for various fixed rate 30y cohorts. Current
should top 34 and 47 CPR pipelines and future defaults should add 13-17.5 CPR to the FNMA 2006-07 6.5s for the next
respectively, in Q1 three years. Near term, buyouts should peak in Q1 10 at about 1.25x the corresponding 1y
CDR shown in Figure 5. For the FNMA 2007 6.5s and 7s, they should top 34 and 47 CPR
respectively, in Q1 10. Also shown are our projected total prepayment speeds for various
mortgage rates. Because FNMA coupons have worse credit quality and higher delinquencies, we
expect them to have slightly better convexity than their FHLMC counterparts, prepaying faster in
an extension environment and slower in a refinancing wave.
Compared with 30y fixed collateral, affordability products such as hybrid ARMs and 10/20
IOs should have even more buyouts, due to a worse credit profile. Aggregate delinquency
rates of 10/20 IOs and hybrid ARMs are fives times as high as those of 30y loans. This suggests a
high buyout rate, although buyouts on these products could be more back-loaded than fixed-
rate loans. For more details, see Prepayment Outlook - GSE Buyouts: Are We There Yet?
November 30, 2009.
21 December 2009 24
Barclays Capital | U.S. Securitized Products Outlook 2010
1. FHA recently tightened its standard for streamlined refinancing. Borrowers can no
longer wrap refinancing costs into the new loan.
2. New rules require that when modifying a loan, the modified interest rate cannot be more
than 50bp higher than the prevailing mortgage rate. This should reduce the incentive for
servicers to buy out loans for modifications.
Underwriting changes should Both changes should reduce the callability of GNMA premiums. More importantly, these are
materially reduce GNMA the first steps that FHA is taking to tighten its underwriting in light of increased scrutiny of
callability in 2010 at its dwindling insurance fund. As HUD Secretary Donovan proposed to Congress, FHA is
likely to make a number of important changes to the FHA program in 2010:
Tighten the requirement for lender qualification, broker supervision, and appraisals
21 December 2009 25
Barclays Capital | U.S. Securitized Products Outlook 2010
Increase quality control, such as reducing maximum permissible seller concessions from
6% to 3%.
Admittedly, the final terms and timing of these changes are still uncertain. Most likely, they will
be implemented after Q1 10, and it will take a few additional months before they take full
effect. Nevertheless, it is highly likely that they will reduce the refinancing efficiency and the
servicer buyouts of GNMA pools, resulting in slower speeds and a better convexity profile in
H2 10.
In contrast, FNMA prepayments should trend higher because its underwriting has already
tightened greatly. In fact, there should be a gradual normalization in conventional lending.
Moreover, GSE buyouts are set to ramp up, further boosting conventional speeds. All these
trends point to a gradual compression in the GNMA/FNMA speed differential over 2010.
Figure 7 shows that in 2009, GNMA premiums prepaid faster than FNMA due to more
efficient refinancing and higher servicer buyouts. For the 2007 6.5s, the differential was
more than 20 CPR. This is likely to change in 2010, as we expect this differential to
compress to 5-10 CPR (Figure 8), although adverse selection in TBA should still lead to a
sizable carry disadvantage for GNMA higher coupons relative to conventionals.
Since this pattern should persist for 2010, we expect conventional OTM speeds to remain
below home sales. If mortgage rates increase significantly (say, to 7%), discount speeds
should drop to 4-5 CPR. At that speed, the FNCL 4-5% coupons would have an 11-12y
average life. In addition, the MBS universe is now more concentrated in lower coupons than
at any other time in history (Figure 10), further increasing extension risk.
Figure 7: GNMA versus FNMA speeds in 2009 Figure 8: Expected GNMA versus FNMA speeds in 2010
CPR (Feb-Nov 09 Report) Involuntary Projected 1y CPR Involuntary
45 Voluntary
45 Voluntary
40 40
35 35
30 30
25 25
20 20
15 15
10 10
GN FN GN FN GN FN GN FN GN FN GN FN GN FN GN FN GN FN GN FN GN FN GN FN
2006 2007 2008 2006 2007 2008 2006 2007 2008 2006 2007 2008
6 6.5 6 6.5
Note: Actual involuntary CPR for GNMA; estimates for FNMA. Source: GNMA, Note: Assuming a 5% no-point mortgage rate. Source: Barclays Capital
FNMA, Barclays Capital
21 December 2009 26
Barclays Capital | U.S. Securitized Products Outlook 2010
The 2009 GNMA production To mitigate this risk, investors could consider 2009 GNMA production: defaults and servicer
should experience similar levels buyouts should put a floor under discount GNMA speeds, offering substantial extension
of delinquencies and buyouts as protection. As a reference, the 2007 origination GNMA 5s, the lowest coupon produced in size
the 2006-07 originations. in that year, have been defaulting at more than 7% per year since origination. Although FHA
asserted that the credit quality of its recent origination has improved, we have found no
evidence that the default rate of the 2009 production will be substantially lower than earlier
vintages. Figure 11 shows the origination FICO of FHA loans, and Figure 12 shows GNMA
cumulative default rate by quarterly originations. Despite a notable increase in the average
FICO of FHA loans since early 2008, default rates of the 2008 and 2009 origination have been
right in line with the 2006-07 production. This implies that delinquencies and servicer buyouts
for the 2009 production should be comparable to the 2006-07 originations.
Figure 9: Discount CPR to be driven by voluntary home sales Figure 10: Migration of the MBS universe to lower coupons
7
40%
6
5 20%
4
0%
3 4.75 5.25 5.75 6.25 6.75 7.25 7.75
Jan-06 Jul-06 Feb-07 Aug-07 Mar-08 Sep-08 WAC
Source: National Association of Realtors, FNMA, Barclays Capital Source: FNMA, Barclays Capital
Figure 11: FNMA versus FHA: FICO by origination month Figure 12: GNMA cumulative defaults by origination quarter
600 0%
Sep-07 Mar-08 Sep-08 Mar-09 Sep-09 3 5 7 9 11 13 15
Source: FNMA, FHA, Barclays Capital Note: defaults are defined as 90+day delinquencies and as of a percentage of
original balance. Source: GNMA, Barclays Capital
21 December 2009 27
Barclays Capital | U.S. Securitized Products Outlook 2010
GNMA credit performance has not improved with higher average FICO for a couple of reasons:
1. Delinquencies are driven not by the average borrower, but by the worst borrowers.
Although sharply tightened conventional lending has driven relatively better borrowers
into the FHA space and lifted the average FICO, many subprime borrowers have also
found their way into FHA. These loans should keep delinquencies high.
2. Extremely high LTV, combined with weak HPA and a high jobless rate, is a perfect
recipe for defaults. Figure 13 shows that throughout the credit crisis, the average
original reported LTV of FHA borrowers has stayed close to 95%. In actuality, it should
be even higher because FHA streamlined refinancing does not require a reappraisal.
Many borrowers who put a 3.5% down-payment to buy a home need stable income
and significantly positive HPA to be able to afford the mortgage. However, right now
neither condition seems to hold, as HPA should remain weak and the jobless rate high
for some time. Consequently, we expect the GNMA delinquency level to stay elevated.
Lower coupon GNMA should All in all, involuntary prepayments should add at least 5-7 CPR to the 2009 GNMA 4.5s and
provide excellent extension 5s. This should give them great extension protection relative to their FNMA counterparts if
protection to their FNMA rates back up significantly, with about a four-year difference in average life (Figure 14).
counterparts.
Figure 13: FNMA versus FHA: Original LTV Figure 14: GNMA to provide substantial extension protection
95 12 11
11.4 11.0 10
90 10
9
8 8.1
7.6
85 8
6.7
6
7
80
4
6
75 2 5
70 0 4
4 4.5 5 4 4.5 5
65 GN FN
Sep-07 Mar-08 Sep-08 Mar-09 Sep-09 Voluntary CPR Involuntary CPR Average Life
Source: FNMA, GNMA, Barclays Capital Note: Assuming a 7% no-point mortgage rate. Source: Barclays Capital
21 December 2009 28
Barclays Capital | U.S. Securitized Products Outlook 2010
AGENCY MBS
There are several ways to short the basis, each with its own advantages and drawbacks
(Figure 1):
Short TBAs – Sell TBAs and hedge the duration in Treasuries/swaps. The main advantage
of using TBAs is liquidity. The TBA market also allows flexibility in terms of choosing the
right coupon to short. The disadvantage is the negative carry.
Long IOs hedged with Treasuries – IOs (trust, excess, or even inverse) benefit from
higher mortgage rates. As Treasuries benefit from lower Treasury rates, this
combination is an implicit spread widener. A key element to this trade involves picking
the right collateral for the IO. We recommend selecting collateral with limited exposure
to non-interest rate factors (buyouts, underwriting changes, etc).
CMM – CMS swaps – This has been an increasingly popular way to short the basis,
especially for non-mortgage investors, owing to the clean execution (no daily re-
hedging is required). And while the trade carries positively, unwinding the trade may not
always be optimal compared with holding it to expiry.
21 December 2009 29
Barclays Capital | U.S. Securitized Products Outlook 2010
Coupon swap hedged with TSY Significant carry Less than perfect correlation
Muted prepays Coupon swap technicals
We believe that IOs hedged with Treasuries are one of the best ways to short the basis now.
This trade has significant carry, even though the correlation is not perfect. IOs still look
cheap to where prepays are coming in and potential tightening in IO spreads should help
this trade. We do not see any signs of significant prepay pick-up in the near term.
Roll richening at the beginning of 2009 was caused by roll purchases by the Fed, to improve
funding in higher coupons. But rolls have continued to trade rich even as the Fed stopped
buying (and sold some rolls in recent months). Rich rolls are the result of two factors: steady
Figure 2: Compare GC, roll funding, and 1m Libor Figure 3: Roll carry versus owning 2008 LLB FN 5.5s (ticks)
2.0 13
1.5
12
1.0
11
0.5
0.0 10
Nov-08 Feb-09 May-09 Aug-09 Nov-09
-0.5 9
-1.0
8
-1.5 Jan-09 Mar-09 May-09 Jul-09 Sep-09 Nov-09
Dollar Roll 2008 LLB 5.5% carry
21 December 2009 30
Barclays Capital | U.S. Securitized Products Outlook 2010
demand from the Fed; and Increased level of cash in the system. As the first effect begins to
fade in Q1 10, rolls should normalize a little. But we expect rolls to trade marginally rich until
the Fed starts draining liquidity in Q3 10. The other issue that has limited the attractiveness of
specified pools is the muted callability of the mortgage universe: limited prepay differential
between specified pools and TBAs. We expect the current muted prepay environment to
continue into H1 10.
In our base case prepay environment, we look at the carry advantage of specified pools under
two roll funding scenarios: rolls normalize at end of Q1; and rolls normalize at end of Q2
(Figure 4). In general, specified pools seem unattractive, given the long break-even times and
high market pay-ups relative to theoretical valuations. But within the specified pool universe,
we believe that the seasoning story looks cheap compared with loan balance collateral. We
see value in 2006 6s: valuations look attractive, and the recommendation is in line with our
view that high-LTV, high-FICO, good credit premiums should offer good value in 2010.
Market
Market Pay-Up % 1y CPR 1y Carry 1y Carry
Pay-Up of Even Projection Difference Break Even Difference Breakeven
Coupon Collateral Price (32s) OAS (%) (32s) (Months) (32s) (Months)
21 December 2009 31
Barclays Capital | U.S. Securitized Products Outlook 2010
Muted voluntary speeds & buyout spike – In addition to slower voluntary prepayments,
we accelerate the timing of delinquency buyouts to be front-loaded. We assume the
entire delinquency pipeline is flushed out in a month, with the GSEs making a policy
decision to buy loans out.
The effect of slower voluntary In the base case scenario, the middle of the coupon stack (5.5s and 6s) appears to be most
speeds should outweigh any fully valued (Figure 5). High dollar prices, coupled with the model’s projection of higher-than-
buyout spike for 6s and 6.5s realized prepayments, hurts these coupons. In the ‘reduced voluntary speeds’ scenario, 6s
through 7s begin to appear very cheap. This is because slower speeds benefit these high
premium securities. Finally, speeding up the timing of delinquency buyouts (third scenario)
also changes valuations. If buyouts are front-loaded, the credit impaired coupons seem much
richer. For example, the OAS for 7s drop 20-30bp in this scenario as their delinquency
pipelines are the greatest. On the other hand, 6s and 6.5s do not suffer that much. TBA 6s and
6.5s only sacrifice about 5-10bp of OAS, suggesting that the impact of slower voluntary
speeds outweighs the potential negative of a buyout spike for 6s and 6.5s.
Figure 5: FN 30y coupon stack Libor OAS valuations across prepayment scenarios (bp)
Reduced Vol &
Coupon Vintage Base Case Reduced Voluntary Buyout Spike
To confirm this intuition about the 6s and 6.5s, we consider a simple example. Take 2007
6.5s. Assume that the delinquency pipeline is 19% and that all these loans would normally
be bought out over the course of one year. The price effect of flushing the delinquency
21 December 2009 32
Barclays Capital | U.S. Securitized Products Outlook 2010
pipeline all at once is 62 cents (6.5% coupon * 19% delinquency * 0.5y duration = 62 cents).
As the spread duration of 2007 6.5s is 3.0 year, this translates to a 22bp reduction in OAS
(62 cents / 3.0yr = 21bp), very similar to the 20bp (25.9bp – 5.4bp = 20bp) effect we
observe in the model (Figure 5). In summary, we believe that 6s and 6.5s offer the best value
in the conventional coupon stack as the benefit from muted voluntary speeds outweigh any
negative effects of a potential buyout spike. If buyouts jump in the near term, this may be an
even better entry point for this trade, as the delinquency pipeline would be flushed out.
Ginnie Mae buyouts have been In order to gauge the relative effect of involuntary speeds on Ginnie Mae and conventional
much greater than for collateral, consider the current status of delinquency pipelines and roll rates between the two
conventional collateral products. Figure 6 shows our estimate of the 90-day+ delinquencies by coupon and vintage for
conventional collateral as well as actual data for Ginnie Mae collateral. We see that the pipeline
of delinquent loans is surprisingly similar between conventional and Ginnie Mae cohorts, even
though servicers have been aggressively buying out seriously delinquent GNMA loans.
Taken alone, the pipeline data do not suggest that Ginnie Mae speeds should be faster going
forward. However, the roll rate is surprisingly different between the two products. For
example, the current to 90-day+ delinquency roll rate for 2007 GN 6s is 16.4%, compared with
only 8.6% for 2007 FN 6s. This suggests that while conventional delinquencies have
accumulated to similar levels as Ginnie Mae collateral, there are significantly more inflows of
delinquent loans in Ginnie Mae pools. Subsequently, overall prepayment speeds in 2010
should remain stronger for Ginnie Mae collateral relative to conventionals.
4
Please see “Trends and Issues,” Securitized Products Weekly, December 4, 2009.
21 December 2009 33
Barclays Capital | U.S. Securitized Products Outlook 2010
Figure 7: Speed difference is exacerbated for TBA pools Figure 8: GN vs FN carry analysis
CPR 3m FNCL GNSF
Coupon Percentile FNMA 30yr GNMA 30yr Diff
5.5% 20% 35.6 48.7 13.1 Cpn CPR BE Carry CPR BE Carry Carry Diff
30% 30.8 43.1 12.2 5 17.5 9.8 18.6 9.1 0.7
40% 26.5 38.9 12.4 5.5 26.5 8.3 38.9 4.9 3.4
6.0% 20% 40.2 60.4 20.2
6 30.6 7.6 46.7 2.6 5.0
30% 35.1 54.3 19.2
6.5 31.2 7.7 52.3 0.9 6.8
40% 30.6 49.9 19.3
6.5% 20% 42.2 70.3 28.1
30% 34.7 65.0 30.3
40% 31.2 58.9 27.7
7.0% 20% 68.6 80.2 11.6
30% 54.3 73.8 19.5
40% 51.3 64.7 13.4
Source: Fannie Mae, Ginnie Mae, Barclays Capital Source: Barclays Capital
While the roll rate data suggest that Ginnie Mae speeds should be faster than conventionals
for a given coupon/vintage, the situation is even more dramatic when looking at TBA pools. In
Figure 7, we look at prepayment speeds for the fastest paying pools, by percentile, in available
float (after removing pledged to CMO and the Fed). For example, the top 40th percentile of GN
6.5% over the past three months was 28 CPR faster than FN 6.5% speeds. This differential in
TBA speeds has important valuation implications. In Figure 8, we calculate the carry in ticks for
Ginnie Mae and Fannie Mae pass-throughs using long-term CPR projections for TBA. These
results suggest than higher coupon Ginnie Mae collateral should give up 3-7 ticks per month
in carry relative to Fannie Mae collateral; thus we recommend being underweight higher
coupon GN/FN swaps.
Increased buyouts lead to To test the resilience of Ginnie Mae collateral, we look at the projected prepayments for GN
shorter average lives for lower and FN collateral in a 200bp sell-off scenario. In such a scenario, we expect new GNMA lower
coupon Ginnie Mae’s coupons to be floored at 9 CPR, while conventional turnover could be as weak as 4-5 CPR.
This translates into 4-5 year shorter average lives for GNMA pass-throughs relative to their
conventional counterparts (Figure 10). For accounts concerned about extension, an
overweight of lower coupon Ginnie Mae pass-throughs (or lower priced CMOs off of most
Ginnie Mae collateral) makes a lot of sense.
21 December 2009 34
Barclays Capital | U.S. Securitized Products Outlook 2010
Figure 9: Lower coupon GNs look attractive to FNs Figure 10: GN pass-throughs extend less in +200bp scenario
Cpn Issuer Price Avg Life LOAS
CPR Average Life(yr)
4 GN 98-16 8.05 16.8
14 11.8 12
4 FN 98-16 9.01 -9.7
12 11
4.5 GN 101-17 6.19 10.1 11.4 11.0 10
7.05 -8.7 10
4.5 FN 101-08+
9
5 GN 104-00 4.23 -3.8 8 7.6 6.7
8
5 FN 103-25 5.01 -14.5 6 8.1 7
4 6
2 5
0 4
4 4.5 5 4 4.5 5
GN FN
Voluntary CPR Involuntary CPR Average Life
Short-duration assets appear rich, but should remain well bid by banks
However, some banks might find the base case duration and average life of GNMA lower
coupons outside of their comfort zone. Another way to manage extension risk is to focus on
short-duration products. But many of these short bonds (hybrid ARMs, short PACs, short
sequentials) have seen both a strong run-up in dollar price and a significant narrowing of
spread in 2009.
Sector Coupon Price 1yr TROR Yield LOAS A/L Price 1y TROR A/L
30y PT 5.00 103.78 3.68 3.93 -25.0 4.4 93.39 -4.89 10.9
15y PT 4.50 103.91 3.47 3.25 -15.0 3.7 96.20 -5.02 5.8
Short SEQ 4.50 104.50 2.18 2.19 -29.7 2.2 99.20 -0.80 4.0
Short SEQ 4.50 103.56 3.19 3.26 -22.4 3.3 96.56 -2.07 4.3
5/1 hybrid 3.00 101.02 3.40 2.62 -6.0 3.5 93.47 -4.06 5.3
5/1 hybrid 3.50 102.38 2.96 2.50 14.7 2.9 95.86 -2.67 4.9
Floater 0.95 100.00 0.98 0.95 -9.6 4.3 95.48 -1.21 6.6
Source: Barclays Capital
Short-duration CMOs should In Figure 11, we survey a selection of typical bank bonds and look at valuations in both the
remain well bid due to demand base case and in a 200bp sell-off scenario. First, all these assets are priced well north of par,
from the banking sector except for the CMO floater. Also, the absolute yield levels for these securities are rather low.
This presents a problem as banks may not want to commit capital if they are bearish on rates.
Finally, all these securities have moderate to significant extension in a sell-off scenario. Despite
the fact that none of the bonds in Figure 11 appear to be “slam dunk” trades, most of these
short-duration bonds are likely to remain at tight spread levels. If we are correct that MBS
demand in 2010 will be fueled in part by the banking sector, these types of cash flows should
continue to see strong demand.
21 December 2009 35
Barclays Capital | U.S. Securitized Products Outlook 2010
Adding IOs to a portfolio helps The current muted prepayment environment also supports this trade. In Figure 13, we
protect against price declines in compare S-curves for the agency MBS coupon stack over two different time periods: the
a large sell-off scenario first half of 2004 and the first half of 2009. As credit remains tight, slower prepays should
both help IO valuations and limit the risk of premium loss from holding onto IOs. Given how
well IOs fit their needs, even banks (which tend to be notoriously wary of IO risk) might
start adding some next year.
Figure 12: MBS plus IOs have a more stable return profile Figure 13: Prepay profiles are much more stable now
4 20% 40
2
15% 30
0
10% 20
-2
-4 5% 10
-6 0%
0
0 50 100 150 200
-100 -50 0 50 100 150 200
Rate Shift (bp)
FNCL 5 07 (LHS) Rate Incentive (bp)
FNCL 5 + 5% IO (LHS)
IO Coupon Returns (RHS,1Yr) 1H 2004 1H 2009
21 December 2009 36
Barclays Capital | U.S. Securitized Products Outlook 2010
RESIDENTIAL CREDIT
The biggest risks for non-agency valuations are unexpected developments on the
Keerthi Raghavan
legislative and modification fronts.
+1 212 412 7947
keerthi.raghavan@barcap.com PPIP, while less relevant, still offers a credible backstop in a down scenario, along
with the bid for re-REMIC seniors. For the most part, real money and hedge funds
that use leverage will likely drive prices.
On the distressed bond supply front, the new NAIC method of calculating capital
requirements will mostly be a non-event for life insurers. Limited potential supply
from P&C insurers should be manageable.
HAMP was not the first government program to help borrowers; it will not be the
last. While the re-default performance of HAMP mods should be marginally better
than pre-HAMP mods, it will likely be bad enough to trigger potentially big changes
or completely new initiatives.
Servicer tiering along modification rates, types, and behavior on advances will
become a dominant theme in 2010.
Introduction
2009 was another roller coaster year for non-agency RMBS. For a change, though, the year
ended on a high note. In the beginning of the year, legislative uncertainties and weak
technicals heightened risk premiums. Myriad possible loan modification plans, an overhang
of foreclosure moratoria, and cramdown threats kept real money investors on the sidelines.
With banks’ balance sheets heavy with unrecognized losses and the prospect of massive
ratings downgrades, distressed hedge funds provided the only underpinning of demand in
early 2009. At the same time, fundamentals continued to deteriorate rapidly. We
recommended staying underinvested at the beginning of 2009, confident that weak
technicals would provide more attractive entry points later in the year (Please see our 2009
Residential Credit Outlook).
21 December 2009 37
Barclays Capital | U.S. Securitized Products Outlook 2010
$260bn in early February downgrades and fears of cramdowns proved the last straw, creating
strong selling pressure and sending prices another 10-15 points lower. With very attractive
valuations, abating cramdown fears and the potential of huge upside from PPIP, we allocated
20% of equity to long trades in our model credit portfolio on March 6 and another 10% a
week later.
After the PPIP announcement later the same month, prices rallied sharply across the RMBS
sectors in the next several weeks. For example, we were able to close our long 07 alt-A FRM
SSNR and mezzanine trades, initiated in early March, with 37% and 67% ROEs in a short
period (for details on the performance of our model portfolio, please see our latest Securitized
Product Weekly). The price action for the second half of the year was driven predominantly by
expectations around non-recourse leverage and the emergence of repo financing, as well as
the strong demand for re-REMICS. The emergence of significant demand from real money
investors compressed unleveraged yields on jumbos to high single digits and yields on alt-a
paper to low double digits.
Despite higher prices, we enter 2010 with a much more benign outlook. We expect home
prices to depreciate only about 8% more from current levels and are likely seeing the first
signs of credit burnout in subprime. While we expect severities to increase further before
levelling off, the probability of an extremely dire scenario is much lower. The overhang of
legislative and modification-related uncertainties remains, but larger risk premiums than
other asset classes compensate for it. The biggest positive, though, comes from the
technical front. With ample liquidity provided by third-party repo lines and non-recourse
government PPIP leverage, demand should be quite strong at least for the first half.
Downgrades are out of the way, and we are unlikely to see anything similar to 1Q09, when
a deluge of forced sellers crowded the market.
Overall, we believe that even with high defaults and loss expectations, current non-agency
valuations are attractive relative to other risky assets. Further evidence of diminished tail risk
and the improved availability of cheap leverage are likely to drive risk premiums lower. As
risk premiums compress, deal selection will become even more critical and result in tiering
across several dimensions. We explore some of these issues and our top trade
recommendations in the rest of the article.
21 December 2009 38
Barclays Capital | U.S. Securitized Products Outlook 2010
While some of these negative performance trends should persist in the first half of 2010, we
expect meaningful improvement starting the second half. In this section, we analyze the
three main drivers of performance – defaults, severities, and prepays – and discuss the main
performance themes that investors should look for in 2010.
Analysis of older vintage subprime collateral can shed some light on how pool
characteristics improve over time as a result of burnout. Figure 2 shows the collateral
composition for 2003 subprime over time, broken out by original FICO. As credit-impaired
borrowers exited the pool, composition shifted toward higher FICO buckets. This changing
composition can have a meaningful effect on overall collateral performance. As Figure 3
shows, performance starts to stabilize as collateral composition improves. Once a
significant proportion of worse credit-quality loans have exited, the loans that remain in the
pool tend to show stable payment behavior. As a result, despite significant home price
depreciation in the past two years, 2003 vintage subprime credit performance has remained
roughly constant (Figure 3).
Figure 2: Composition shifts across FICO buckets Figure 3: 60+ delinquency rate by wala, ‘03 vintage subprime
Factor 60+ (%
Factor 550- 600- 650-
Curbal)
Month <550 600 650 700 >700 120% 18
16
Oct-03 15.7% 23.1% 31.0% 19.4% 10.8% 100%
14
80% 12
Oct-05 14.8% 21.3% 29.6% 20.0% 14.3%
Factor 10
60%
Oct-07 14.2% 18.8% 27.4% 21.2% 18.3% 60+ (% Current Balance) 8
40% 6
Oct-09 14.2% 18.6% 27.4% 21.5% 18.3% 4
20%
2
0% 0
1 6 11 16 21 26 31 36 41 46 51 56 61 66 71
Note: Data shown for 2003 subprime. Source: LoanPerformance, Barclays Capital Note: Data shown for 2003 subprime. Source: LoanPerformance, Barclays Capital
21 December 2009 39
Barclays Capital | U.S. Securitized Products Outlook 2010
Figure 4: Sample CDRs in base and burnout scenario Figure 5: Current-to-delinquent roll rates across pay history
and modification buckets
CDR C to Q
Clean Unmodified Clean Modified
Current Transition Rates Burnout scenario 16%
18%
Dirty Unmodified Dirty Modified
14%
17%
12%
16% 10%
15% 8%
6%
14%
4%
13%
2%
12% 0%
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 Mar-09 May-09 Jul-09 Sep-09 Nov-09
Month Forward
Note: Sample CDR curves shown for 2007 subprime. Source: LoanPerformance, Note: Clean loans refer to loans with no delinquency history since Jan 2008. Dirty
Barclays Capital loans have been 60 days delinquent at least once. Data shown for 2006
subprime. Source: LoanPerformance, Barclays Capital
As Figure 4 shows, CDRs for the two scenarios are not very different for the first six months
as loans already in 60+day delinquency work their way through liquidation. The effect of
credit burnout is more pronounced later, when current-to-delinquent roll rates fall gradually
as burnout kicks in. This would also be accompanied by somewhat faster voluntary prepays,
resulting in lower defaults.
21 December 2009 40
Barclays Capital | U.S. Securitized Products Outlook 2010
Figure 6: Re-default rates across payment reduction buckets Figure 7: Debt to Income ratios for permanent HAMP mods
Cum
Mean DTI
Delinquency
100% 90
90% 76.4
80
80% 70 24.6
70% 60 51.8
47.2
60% 50 16.1
50% 0-10 40 31.1
40% 10-20 30
30% 20-30 20
20% 30-40 10
10% 40-50
0
0%
Pre mod Post mod Pre mod Post mod
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 Back DTI Back DTI Front DTI Front DTI
Months since modification
Note Data shown for subprime loans modified in 3Q08. Source: Loan Source: Congressional Oversight Panel October Report, Barclays Capital
Performance, Barclays Capital
Re-default performance for loans modified in Q3 08 has been dismal, with more than 60%
relapsing into deep delinquency already. However, HAMP has been more aggressive than
earlier mods – reducing borrower payments by 30-40%, compared with earlier modification
efforts that typically reduced monthly payments by 15-20%. As Figure 6 shows, higher
payment reductions reduce re-default rates, but only by 5-10% for that magnitude of
payment change.
In addition, the Congressional Oversight Panel report released in October has some data
that indicate that HAMP loans may perform better than expected for other reasons. HAMP
modified borrowers seem to be benefiting from a reduction in debt other than their first-lien
mortgages – possibly from mandated HUD counseling for borrowers with back debt-to-
income ratios (DTIs) above 55. As Figure 7 shows, the back DTIs of modified borrowers fell
more than their front DTIs, implying debt reduction outside of the first mortgage. While
there are other reasons for back DTIs to fall, such as a chapter 7 bankruptcy, any
counselling related benefits are unique to HAMP and may help improve HAMP re-default
performance vs. prior experience.
On the flip side, HAMP does not address the issue of negative equity, which is one of the
primary drivers behind default. Also, HAMP-related modifications do little to ease the
payment burden on more than half the modified borrowers who have very high back DTIs.
…but HAMP may fare (Please see our article on HAMP re-defaults in the Securitized Products Weekly October 30,
a little better 2009) Taking these factors into account, we expect overall HAMP re-default rates to show
not more than a 10-20% improvement over the default rates seen in past mods. The large
number of modifications will also serve to back-end the CDR curve, and default models
need to take this into consideration since it has significant redistributive effects on
valuations.
21 December 2009 41
Barclays Capital | U.S. Securitized Products Outlook 2010
Servicers can also influence deal severities through P&I advances. Government-imposed
foreclosure moratoria, combined with modification programs, have extended foreclosure/
REO timelines, especially for subprime. With timelines stretching out for more than 20
months, servicers are becoming more aggressive in stopping advances on loans that they
believe have little or no chance of recovery.
Severities should improve in late We expect home prices to decline in the first half of 2010 before stabilizing and growing at a
2010 with rising home prices fairly slow pace. (For more details, please see “Housing risks and prospects”). Worsening
mark-to-market loan-to-value ratios (LTVs) in the near term due to adverse home prices
should cause severities to rise in the first half of 2010. However, since large-scale
modifications would delay defaults, most redefaults will occur when home prices stabilize
somewhat. This would serve to reduce losses on the pool compared with a no-modification
scenario. In addition, as the share of modified loans increases, the headline severity number
40
55
35
30
45
25
20 35
15
10 25
Oct-08 Jan-09 Apr-09 Jul-09 Oct-09 Oct-08 Jan-09 Apr-09 Jul-09 Oct-09
Note: Data shown for 2006 vintage and 130-180 MTM LTV. Source: Note: Data shown for 2006 vintage subprime and 130-180 MTMLTV. Source:
LoanPerformance, Barclays Capital LoanPerformance, Barclays Capital
21 December 2009 42
Barclays Capital | U.S. Securitized Products Outlook 2010
will also get a boost from lower P&I advances due to lower average monthly payments. Please
see Why are first-lien severities falling? for more details on our severity projections for 2010.
We expect FHA to accelerate the tightening of its underwriting standards in light of the poor
recent performance of its collateral. While we might see some relaxation in GSE guidelines
6-18 months out, they will still be tight enough to exclude worse-quality, high-LTV non-
agency borrowers. (Please see “Agency underwriting: When will the squeeze ease?” for
further details). Absent loosening, voluntary prepays will remain slow, as a lot of non-
agency borrowers eligible for these programs have already prepaid. On the new origination
front, we believe it is too early to start talking about subprime/alt-A originations. In light of
these factors, we don’t expect any moves from the current sluggish speeds in subprime and
option ARMs in 2010. However, longer duration 2007 subprime bonds could gain even with
modest increases in voluntary prepays in 2011 and beyond.
…but new prime originations As for prime collateral, the high percentage of the refi-eligible population in the jumbo fixed
could increase jumbo CRRs sector meant that borrowers were able to take advantage of lower rates to refinance into
agency-backed loans. As the refi-eligible population prepaid out toward the end of
September accompanied by higher mortgage rates, voluntary speeds fell sharply, by 5-8
CRR (Figure 10). Despite regulations mandating that the originator hold some economic
interest in new originations and potential consolidation issues for balance sheets, there is a
case for some new non-agency origination in the prime sector Overall, we expect voluntary
prepays for prime borrowers to range between 10-12 CRR in 2010, alt A in the mid-single
digits, while subprime and option ARM speeds should remain in the very low single digits.
If rates remain low, deal-level tiering in terms of refi eligibility will gain importance. Even
now, there is a considerable difference in prepay speeds between refi-eligible and non-
eligible populations. This can be seen clearly in Figure 11, which breaks out all 2007 vintage
Figure 10: Refi-eligibility and prepay speeds over time Figure 11: Deal CRRs across %refi-eligibility quartiles
% refi- % refi-
CRR CRR
eligible Refi-eligible CRR eligible
16% 25% 25% CRR % Refi-eligible 30%
14% 25%
20% 20%
12%
20%
10% 15%
15%
15%
8%
10%
6% 10% 10%
4% 5% 5%
5%
2% 0% 0%
0% 0% 1 2 3 4
A pr-09 M ay-09 Jun-09 Jul-09 A ug-09 Sep-09 Oct-09 No v-09 Quartiles (by % refi-eligible)
Note: Data shown for 2007 jumbo fixed collateral. Source: LoanPerformance, Note: Data shown for 2007 jumbo fixed collateral for November remits. Source:
Barclays Capital LoanPerformance, Barclays Capital
21 December 2009 43
Barclays Capital | U.S. Securitized Products Outlook 2010
jumbo fixed deals into four quartiles based on their refi eligibility. The top quartile of deals
prepaid 8 CRR above the bottom quartile in the most recent episode.
Loss expectations
Overall, we expect performance to worsen in the first half of 2010 as home prices decline,
timelines continue to lengthen, and severities rise. We expect the effects of more aggressive
modifications, credit burnout, and recovering home prices to kick in toward the second half
of 2010, leading to some improvement in performance. Our loss expectations for the base
HPA, assuming an expected level of HAMP mods, are shown in Figure 12. However, given
the poor conversion rate for HAMP trial mods and likely poor performance, we are likely to
see significant changes to the HAMP program or completely new initiatives.
Figure 12: 2007 vintage loss projections (home prices down 10% forward, nationally)
Forward % current
Lifetime loss
default loans Forward
Sector Loan type expectation
expectation defaulting (% severity (%)
(% origbal)
(% curbal) curbal)
Fixed 11 37 33 41
Jumbo
Hybrid 24 57 52 48
Fixed 26 66 59 45
Alt-A
Hybrid 37 74 65 58
Negam Overall 54 86 80 68
Alt-B Overall 47 84 76 64
Subprime Overall 53 90 84 69
Note: Shown for a scenario of future 10% drop in national Case-Shiller home prices for 2007 vintage. Source: Barclays
Capital, Loan Performance
21 December 2009 44
Barclays Capital | U.S. Securitized Products Outlook 2010
However, significant problems plague the program in its current form. We discussed the
issue of high re-default rates in detail in the previous section. This remains a major problem
with the program and will likely become even more apparent in 2010 as we get more data
on HAMP redefaults. That, however, is not the only issue.
Servicer-related obstacles
Servicer economic It is well known that capacity issues are a drag on modifications. In addition, the rate of
considerations and legal liability modification has been lower for securitized non-agency loans compared with GSE or
concerns also limit HAMP portfolio loans. This indicates that issues related to servicers’ liability to investors have still
not been sorted out. Finally, servicers may also resist modifying balances because it reduces
their servicing strip. They would prefer to reduce rates and extend term, which ends up
being not as effective as balance reduction in many cases. As timelines get stretched
21 December 2009 45
Barclays Capital | U.S. Securitized Products Outlook 2010
because of failed modifications and short sales, the number of advances that servicers need
to make will also increase.
200
160
120
80
40
0
May-09 Jun-09 Jul-09 Aug-09 Sep-09 Oct-09 Nov-09
Source: US Treasury, Barclays Capital
We first take a look at the different changes that are likely given the above scenario. Most of
these changes target loan modification, while a few target foreclosure timelines and REO
stock. Not all the changes are negative for non-agency securities. Some, like a revised H4H or
stronger second-lien modification program, can actually be beneficial for first-lien valuations.
HAMP-specific changes
Changes to HAMP would Setting DTI target below 31 or targeting back DTI: Early experience with HAMP has
increase eligibility and shown that even 31 DTI is too high for many borrowers. A lot of these borrowers have
performance second liens, credit card, or medical expense-related debt among others. These
borrowers may not have enough money to make payments even after the payment on
their first liens is reduced to 31 DTI and will likely need more assistance. In addition,
many borrowers are left out of the ambit of HAMP for having DTI lower than 31 but they
are also increasingly vulnerable to defaults. These changes could increase eligibility,
improve trial mod to permanent mod conversion, and reduce re-defaults.
21 December 2009 46
Barclays Capital | U.S. Securitized Products Outlook 2010
Bankruptcy cramdowns
Outside chance of cramdown Bankruptcy cramdowns, although unpopular with investors, have some political backing as
provision exists a way to reduce borrower indebtedness through the judicial route. Ever since it was voted
down in the Senate, it has been used as an occasional threat to push servicers to carry out
more modifications. If there are signs that the HAMP program is faltering or an insufficient
number of permanent modifications are being carried out, then there could be a renewed
push to institute this procedure. While the potential negative effect of cramdowns is
significant, we believe that the chances of its enactment remain minimal.
Foreclosure moratoria
Foreclosure moratoria may There have been sporadic instances of foreclosure moratoria in the past year. However, as
become politically attractive election season nears, the ramifications of a large number of borrowers being foreclosed on
will be large and may result in the imposition of more widespread foreclosure moratoria.
This would extend timelines and be a net negative for valuations up the capital structure.
Foreclosure moratoria could also be used as a bargaining tool to get servicers to sign up for
HAMP or any new programs.
21 December 2009 47
Barclays Capital | U.S. Securitized Products Outlook 2010
… While the prime sector is likely More modifications in the prime market: The first wave of modifications has mostly been
to see more mods in subprime mortgages as more of those borrowers have been as risk of or in default.
However, successful permanent modifications are very difficult in that space because the
borrowers are more credit constrained and have poor pay histories. They are also unlikely to
have well documented and steady sources of income. As a result, there may be a push
towards more prime modifications just to hit the target number of modifications numbers.
21 December 2009 48
Barclays Capital | U.S. Securitized Products Outlook 2010
participants that sold at the beginning of the year started building back their positions after
prices bottomed in Q1 2009.
Figure 15: Outstanding non-agency bonds and holders of original AAAs, $bn notional
End Q3
Investor Class
2008 2009
Secondary market supply likely muted versus the deluge in early 2009
Technical conditions going into Late 2008 and early 2009 saw some of the most panicked trading sessions in the non-
2010 are a complete turnaround agency market, with little two-way interest. This was a period in which overseas investors,
from last year… money managers, insurance companies, and some banks were dumping non-agency
securities into an extremely difficult market in terms of both default/loss expectations and
risk premiums. The prospect of rating downgrades, especially on alt-A and prime securities,
and the consequent increase in capital requirements at banks and insurance companies
exacerbated the situation. We enter 2010 after an almost complete turnaround in the above
technical factors. The new insurance company regulatory capital regime could prove to be a
problem, but overall, we still expect the risk of a deluge of supply hitting the market to be
small for a couple of reasons.
…with rating downgrades an First, non-agency holders are much less likely to have to dump securities in 2010. Rating
issue of the past… downgrades are no longer an issue, with a large proportion of original 2007 vintage AAA
securities at or below the CCC level (Figure 16). Of bonds originally rated AAAs, about 70-
90% of various sectors have been downgraded to BBB or below. Some bonds could be
downgraded from BBB to CCC, but the effect of these downgrades is expected to be much
smaller than the AAA to BBB downgrades last year. While older vintages have more
outstanding AAAs, and could have more downgrades, we expect the effect to be relatively
muted at this point.
Figure 16: 2005-2007 vintage current worst rating compositions for original AAAs
BBB to CCC and
Sector Product AAA AA A single B below
21 December 2009 49
Barclays Capital | U.S. Securitized Products Outlook 2010
Banks have strong capital Commercial banks: After the stress test and TARP fundings, banks no longer have major
positions after stress tests capital issues and are not forced sellers. Furthermore, some of largest banks have ring-fenced
their most toxic RMBS securities away from their balance sheets, giving them freedom from
mark-to-market issues. Also, banks tend to shift their portfolios to securities from loans after
recessions. While this shift is likely primarily to increase agency MBS holdings, there could be
some interest in the non-agency space as well, given the elevated yields.
GSEs are in runoff mode and GSEs/FHLBs: While GSEs are required to shrink their portfolios by about $160bn in 2010, we
unlikely to sell think that the portfolio runoff, which is averaging $20bn a month, should be adequate to
bring about the decrease. Even if rates back up and prepays slow a lot, most of the selling is
likely to come from their agency portfolio and not the non-agency book. Likewise, FHLBs
are also not expected to be sellers of non-agency paper, despite potential capital issues.
Money managers will be net Money managers: Over the second half of 2009, money managers became net buyers of
buyers in 2010 non-agency paper. This should continue into 2010, especially through the PPIP. While PPIP
might not expand to its proposed size of $40bn in buying power, it will remain a backstop to
the market. Furthermore, if prices do fall, PPIP managers might find it a lot easier to raise
money and reach the $40bn mark, which should have a supporting effect on prices.
Insurance companies are the largest source of uncertainty from the point of view of
potential secondary market supply, in large part because of the new method replacing the
existing rating-based capital requirements for statutory accounting. The new method uses a
third-party credit model (with PIMCO as the vendor) to come up with an intrinsic value for
each RMBS security. This intrinsic value is used to calculate a maximum price for each NAIC
capital requirement level (Figure 17). The capital and mark-to-market requirements are
then calculated based on the level of the amortized cost basis of the security in relation to
the maximum price levels for each of the NAIC capital levels. We examine this process in
more detail below.
Figure 17: NAIC charges and maximum price breakpoints for a $76 intrinsic value bond
Life insurer P&C and health insurer
NAIC designation RBC charge Midpoint loss Price breakpoint NAIC designation RBC charge Midpoint loss Price breakpoint
21 December 2009 50
Barclays Capital | U.S. Securitized Products Outlook 2010
The NAIC initial designation is identified by finding the highest numbered bucket for which
the maximum price is above the amortized cost. So, as Figure 17 shows, this bond would
fall under designation 4 for life insurers and designation 5 for P&C insurers. For bonds falling
into designation 6 for life insurers and designations 3-6 for P&C insurers, the new
accounting requirements force the insurer to hold the bond at fair value; for other
designations, the bond can be held at its amortized cost basis.
As the above example shows, depending on the portfolio and expected loss, as well as the
type of insurer, we could see radically different capital charges and markdowns for various
insurance companies. For life insurers, in general, as long as the intrinsic value is above
$73.5, the bonds should have a designation of 5 or lower (assuming that the amortized cost
is at or below $100) and would require only capital charges but not mark-to-market hits.
For P&C insurers, on the other hand, since fair value is used for designation 3 or below, a
bond with an amortized cost basis of par can have an intrinsic value of no less than $98.5 to
continue to use an amortized cost basis.
...with P&C insurers at the While the immediate capital effect of either selling the bond at fair value or holding the bond
biggest risk at fair value with a 0.3% capital charge is small, we could see some selling, especially if the
insurers want to get rid of any future mark-to-market volatility. As we pointed out above,
this risk is higher from P&C insurers, which hold about 22% of the RMBS owned by all
insurers. This translates to about $37bn notional of RMBS AAAs, which is a substantial
number in relation to trading volumes. However, insurers might explore other potential
avenues apart from outright selling, such as portfolio re-REMICs or other risk transfer trades
that could be structured at a lower overall cost while keeping the upside.
21 December 2009 51
Barclays Capital | U.S. Securitized Products Outlook 2010
4,000
3,500
3,000
2,500
2,000
1,500
1,000
Oct-04 Oct-05 Oct-06 Oct-07 Oct-08 Oct-09
MMKT funds Long Term Bond Funds
Source: Haver, ICI, Barclays Capital
21 December 2009 52
Barclays Capital | U.S. Securitized Products Outlook 2010
First, there are still sectors, such as alt-A hybrid SSNRs, that offer attractive PPIP yields of
18-20%. Second, even in sectors in which prices are too high to earn the required PPIP
yields, PPIPs will remain a good backstop bid should prices go down. Finally, while the PPIPs
have raised only about $5bn in private capital so far out of the maximum possible $10bn,
we believe that if there is any substantial widening in risk premiums from here, it will turn
into an opportunity for the PPIPs to raise more money and support the market. This, along
with the ability of the government to expand the program if required, should keep levels of
non-agency prices fairly firm and not substantially below where they trade at the moment.
Other factors
…with re-remics lending a Re-REMIC issuance will likely be another source of support for prices in the non-agency
helping hand sector. While re-remics could become impractical if unleveraged yields drop much further in
the jumbo market, there is still some leeway in the alt-A sector. Currently, Fitch and S&P
have refused to rate any re-remic bonds from the alt-A sector as AAAs but we could still see
some demand for non-AAA rated re-remic seniors. This, along with the PPIP, should
provide a support mechanism for prices.
Primex (or ABX.Prime): The dealer community recently voted to start trading an ABX-like
index for clean jumbo/prime passthrough AAAs in 2010 called the ABX.Prime (or Primex).
Although final selection criteria and structuring details for PrimeX have not been released,
we think that investor concerns about cash prices’ plummeting upon PrimeX’s rollout are
overblown. While PrimeX provides a way to short prime mortgages, we think attractive
unleveraged yields and even higher leverage might, if anything, make it a preferred avenue
for putting on long recovery trades. However, PrimeX could amplify and accelerate the
effects of any sector-wide blowout in credit.
21 December 2009 53
Barclays Capital | U.S. Securitized Products Outlook 2010
In summary, we expect technicals in the secondary market to remain fairly strong through
at least H1 2010. There are unavoidable concerns about government programs and
performance that, if they materialize, could change the secondary market dynamics
substantially. But on balance, we enter 2010 with strong technicals to cope with these risks.
21 December 2009 54
Barclays Capital | U.S. Securitized Products Outlook 2010
Relative value
Portfolio positioning for 2010
No easy trades in 2010, but we After a year in which almost all non-agency SSNRs have increased 20-30% in price, 2010
expect risk premiums to will likely be the year in which tiering returns to the sector in a big way and security
compress further… selection start to take precedence over prevailing trends. While most bonds were priced at
high enough yields at the beginning of 2009 and going long in March would have been the
best trade, we believe that 2010 is unlikely to offer up such rewarding trades. Overall prices
continue to reflect fairly dire base-case scenarios, but the probabilities assigned to extreme
scenarios by the market have been declining. As a result, risk premiums in the sector have
compressed dramatically, in line with other risky asset classes. While there are risks from
modification-related uncertainty or misdirected government actions, strong technical
positives should compress yields further from here. While the path over the year may not be
smooth and we will watch vigilantly for any missteps/unintended consequences from the
government-directed programs, in our base-case expectation, we still expect prices to be
higher a year hence. As a result, we recommend being long outright in certain areas, such
as alt-A hybrid SSNRs and long jumbo fixed SSNRs with repo leverage.
…as the availability and level of The return of leverage, both from the prospect of PPIP funds entering the fray and from the
leverage improves comeback of the third-party repo market, hastened risk premium compression in 2009. This
came about as the investor mix shifted from distressed hedge funds buying unleveraged to
real money investors and hedge funds buying with financing. We expect this theme to
continue into 2010 and see no reason the market should continue to trade at 8-10%
unleveraged returns, especially if the concerns about performance and government
intervention in modifications prove to be overblown. Unleveraged yields across different
asset classes (Figure 21) still suggest that RMBS remains attractive compared with other
risky sectors. As investors get more comfortable with loss/modification assumptions, we
are likely to see more leverage and, consequently, lower unleveraged yields.
Security selection will be the As the market moves to this lower yield environment, security selection will become
overriding theme for valuations increasingly important, since bonds that are even a percent apart in yields could, with leverage,
produce yields that are potentially 10-15% different. Also, the cushion that investors had with
the high risk premiums in 2009 enabled them to be more flexible in security selection, with
selection based on broad categories. This cushion has partially melted away and will continue
to do so in 2010. This should also increase the importance of security selection, as investors
are less tolerant of losses/performance worsening at these low yields.
21 December 2009 55
Barclays Capital | U.S. Securitized Products Outlook 2010
Trade recommendations
Cross-sector relative value: Favor alt-A hybrid SSNRs and jumbo hybrids leveraged
While security selection will be important in 2010, there are still some outright long
opportunities in the non-agency sector. Figure 22 shows our projections for yields on
various non-agency SSNR AAA bonds. We look at yields under three scenarios, a base case
(10% decline in national HPA), stress case (20% decline in national HPA), and a recovery
case (0% drop in national HPA). Our scenarios also assume that about 35-50% of
delinquent loans will be modified across various sectors, with a re-default rate of ranging
from 45% to 75% across sectors and type of modification. We also look at three kinds of
loss-adjusted yields: unleveraged cash bond yields, leveraged yields for PPIP equity investors
assuming a full turn of leverage, and leveraged yields for investors using the REPO funding
described in Figure 22.
We favor alt-A hybrid SSNRs for Overall, we like the alt-A hybrid SSNRs on an outright unleveraged basis with yields of 10-
their high unleveraged yield and 13%. We believe that the IO reset concerns related to alt-A hybrids are exaggerated. We see
current pay nature of cash flows the reset risk as more of a modification risk, since we believe that most loans facing a
payment reset will be modified at least to keep their payments the same as they were before
the conversion from IO to P&I payments. This modification implies about a 2% rate
reduction, which is similar to the payment reduction assumed in the model scenarios below.
We also believe that PPIPs will end up preferring alt-A hybrid WAC PT SSNRs for their high
coupon and current-pay characteristics. The outright unleveraged yields also mean that we
like these bonds from the PPIP and REPO-leveraged yield perspectives. For more details on
why we believe this is an ideal bond for PPIPs, please see What will PPIP funds buy?
Jumbo fixed leveraged bonds are Given our view that REPO leverage costs will decline over 2010 and haircuts fall, we like
most likely to see financing- jumbo/alt-A fixed-rate SSNRs bought with REPO leverage. While other hybrid SSNRs look
related gains and are relatively more attractive on an outright yield basis, we think that the likelihood of haircut tightening
safe from losses is highest for the safest of bonds, i.e., the jumbo fixed SSNRs, and they provide 15-30%
leveraged yields, which could go up even further with higher leverage.
Figure 22: Unleveraged and leveraged loss-adjusted yields of SNR AAAs, 2007 Vintage
Unleveraged yields (%) PPIP leveraged yields (%) REPO leveraged yields (%)
Sector Product Tranche Price Stress Base Recovery Stress Base Recovery Stress Base Recovery
Jumbo FIX SSNR 84 6.8% 7.7% 8.3% 8.0% 9.8% 10.7% 20% 29% 33%
ARM SSNR 73 5.7% 8.2% 9.5% 6.9% 11.0% 13.0% 12% 27% 34%
Alt-A FIX SSNR 72 7.8% 9.5% 10.3% 10.1% 12.7% 13.7% 18% 26% 30%
ARM SSNR 61 8.3% 11.8% 13.5% 11.7% 17.0% 19.2% 16% 27% 32%
Negam NEG SSNR 49 4.9% 10.5% 12.8% 5.8% 14.5% 17.6% 5% 20% 26%
Subprime MIX PAAA 35 -0.1% 8.4% 12.5% -1.6% 10.6% 16.1% -6.6% 13.2% 22.8%
MIX AAA 32 1.1% 9.3% 13.2% 0.0% 11.8% 17.0% -4.0% 15.4% 24.3%
Source: Barclays Capital
Note: REPO leveraged yields assume the haircuts and funding costs shown in Figure 19. We also assume that forward unleveraged yields remain the same as today’s
yields and a monthly rolling of the financing. Base scenario assumes that home prices fall 10% from here nationally. Stress scenario assumes that home prices fall
30% from here nationally. Recovery scenario assumes no further drop in home prices nationally. All scenarios assume an expected level of modifications suitable for
each housing scenario.
21 December 2009 56
Barclays Capital | U.S. Securitized Products Outlook 2010
Recovery trades:
Subprime LCF, option ARM SSNRs, and alt-A hybrid mezzanine AAAs
Subprime LCF AAAs and option Our base-case expectation for housing is that it will fall another 8% from current levels
ARM SSNRs are most leveraged nationally, which is close to the base-case scenario shown in the yield tables above. This
to a housing recovery… view is elaborated in “Housing risks and prospects.” If, however, housing is much better
than our expectations, bonds such as subprime LCFs and option ARM super seniors have
the most leverage to this recovery. As can be seen from the unleveraged yields in Figure 22,
while their yields do not match up to the alt-A hybrids in the base case, going from a base
to a recovery case, these bonds improve in yields by 2.5-4.0%, versus 1.0-1.5% for the other
SSNRs. Both these bonds are floaters and are much more leveraged to the PO component,
given that their IO component is expected to be worth a much smaller fraction of the total
price than for the WAC PT bonds. Hence, any additional dollar of principal coming in from
higher voluntary prepays, lower defaults, or lower severities as a result of a housing
recovery would benefit these bonds more on a dollar price basis than it benefits the WAC PT
SSNRs, which also end up losing some IO value. As a result, these bonds are a good way to
make a leveraged investment in a housing recovery. There is some incremental benefit from
leveraging these bonds through the repo market, but this works best for lower dollar-price
bonds with enough cash to pay the debt interest while still having only limited downside.
…as are alt-A hybrid Similarly, alt-A hybrid mezzanine AAAs are also leveraged to faster prepay/lower recoveries
mezzanine bonds and, hence, to a housing recovery and offer attractive yields in a recovery scenario, as
shown in Figure 23.
Figure 23: Unleveraged loss-adjusted yields for mezzanine AAAs, 2007 vintage
Sector Product Tranche Price Stress Base Recovery
Tiering trades
In its current form, HAMP relies on individual servicers to screen borrowers for
modifications, conduct NPV analyses, offer trial modifications, and eventually convert trials
into permanent mods. Variations in how servicers perform these functions can lead to
meaningful tiering opportunities between deals. In addition, servicers can also affect deal
cash flows through their own mod programs, short sales, and P&I advances. In this section,
we discuss some possible tiering trades based on servicer behavior
21 December 2009 57
Barclays Capital | U.S. Securitized Products Outlook 2010
Under pressure from the administration, servicers may modify a large number of delinquent
loans in a short period of time and recoup their advances from the pool. This will lead
to principal payments from the pool being diverted from the front cash flow to pay the
servicer while the corresponding losses hit the subordinates. Currently, subprime
60+deliquencies vary between 40% and 50%, and typical delinquency pipelines can range
from 6 to 18 months.
That could affect the price of front cash flows negatively in two ways. First, the bond will
become longer as principal payments are reduced. Second, subordinates will be written
down faster, leading to early crossover in subprime. To show the effect on prices, we
construct an extreme scenario in which the servicer modifies all delinquent loans in the deal
(50% of current balance) over a 3-month period and recoups all advances due from deal
cash flows (nine months of advances). In the second scenario, the servicer recapitalizes the
loans but waits until liquidation to recoup advances. Figure 24 shows bond factors across
time for the front pay, as well as the LCF, in these two scenarios. As we can see in the
example below, the front cash flow bond can lose as much as 10% of the notional, or over
10c on the dollar in such scenario. Even if we assume a more realistic scenario in which
servicers modify only a portion of the delinquent loans in the pool and recoup advances
over 6-8 months, prices could fall by 4-5pts on the first cash flow. We believe this a risk that
investors should keep in mind as modifications pick up steam in 2010.
Figure 24: Subprime FCF and LCF performance in two advancing scenarios, over time
Factor
Crossover occurs sooner when
120% servicer recoups advances
100%
Servicer
80% modifies and
recoups P&I
LCF bond
60%
0%
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42
21 December 2009 58
Barclays Capital | U.S. Securitized Products Outlook 2010
Deals in which servicers push through a higher number of modifications would experience
lower CDRs on the pool – as more delinquent loans are modified instead of being flushed
through REO or short sales. Even with a 50-60% re-default rate, the overall performance in
the higher modification scenario would still be better than in the alternate case, in which
close to 90% of delinquent loans eventually default. As Figure 25 shows, CDRs could vary
significantly between two servicers – simply based on the rate at which they modify loans.
Higher mods would also push re-defaults further into the future, when we expect severities
to be lower than current levels. This would further reduce losses on the deal.
Reduced and more back-ended default curves should increase valuations across the capital
structure. We estimate prices on front and second CF subprime bonds to be about 4pts
(Figure 29) higher for deals in which servicers push through a higher number of mods
(assuming 10% yield).
More principal
3) Share of debt forbearance/forgiveness modifications: We have also seen a
modifications will hasten
considerable variation between servicers in implementing debt forbearance/forgiveness
crossover on subprime
modifications. Servicers have opposing interests in pushing principal mods. While re-default
performance for principal forgiveness modifications has been better than average, servicers
could also stand to lose their servicing strip on the modified loan to the extent that the
balance is forgiven. As Figure 26 shows, while rate reduction mods dominate, a few
servicers, such as Avelo, Fremont, and SPS, have been actively pursuing principal
forgiveness mods.
A greater share of debt forbearance mods would lead to upfront losses on the pool, in turn
leading to higher initial CDRs. However, since debt forgiveness mods typically perform
better than comparable rate reduction mods, re-default rates would be lower (Figure 27).
Higher losses upfront on the forgiven amount would imply that subordinates would be
written down faster on subprime deals, causing crossover to occur sooner. This would
Figure:25: Sample CDR curves across modification scenarios Figure 26: Type of modification by servicer
CDR
35 Servicer %Modified %RR %DF %DF+RR
30
Ocwen 34.0% 89% 2% 9%
25 Fremont 32.6% 84% 1% 15%
20 Indymac 22.7% 92% 0% 7%
Note: High mod/ Low mod scenarios assume that 70% and 15% of delinquent Note: Data shown for subprime loans for original servicer as reported in Loan
loans get modified respectively. Source: Barclays Capital Performance. Source: LoanPerformance, Barclays Capital.
21 December 2009 59
Barclays Capital | U.S. Securitized Products Outlook 2010
Figure 27: Sample CDR curves across debt forgiveness Figure 28: Number of advances by servicer
scenarios
CDR Months of
30 advances
30
25
25
20
20
15
15
10 Low Debt Forgiveness
10
High Debt Forgiveness
5 5
0 0
B o fA Wells IndyM ac JP M NatCity Fremo nt OptOne Ocwen
1 6 11 16 21 26 31 36 41 46 51 56 61 66 71
Month Forward Advances paid Advances missed
Note: High debt forgiveness scenario assumes that 50% of modified loans Note: The data shown is for 2006 Subprime vintage for servicers as reported by
receive principal forgiveness. Low scenario assumes 0%. Source: LoanPerformance. Source: LoanPerformance, Barclays Capital
LoanPerformance, Barclays Capital
benefit the second and third cash flows at the expense of the first cash flow bond as the
principal waterfall switches from sequential to pro rata. On deals in which crossover does
not occur – such as alt-A hybrids – higher debt forbearance should benefit the SSNR as the
mezzanine AAAs get written off sooner. However, the mezzanine bond would take a big hit
as the IO payment is cut short because of the initial losses from debt forgiveness mods. As
Figure 29 shows, doing more or less mods can have a nearly 10% effect on bond prices.
4) Share of short sales: Deals in which servicers move aggressively in pushing short sales
could face lower severities on average. In addition, defaults will be more front-loaded as
timelines to liquidation shrink. Using sample CDR and severity curves for a 10% higher short
sale percentage, we find that prices could improve up to 3-5pts for alt-A SSNRs (Figure 29)
and subprime front cash flow bonds in deals in which servicers liquidate a larger share of
loans through short sales. Severities may also improve for deals in which servicers stop
advances sooner and have lesser P&I to recoup. As we can see in Fig 28, the trend is more
prevalent among smaller servicers that might face higher cost of funding advances.
High
Low High Low debt High debt short
Sector Tranche Base mods mods forbearance forbearance sales
Subprime FCF 55 51 57 56 51 60
Subprime 2nd CF 37 34 39 37 39 41
Alt-A ARM SSNR 60 58 61 60 60 63
Alt-A ARM Mezz 9 9 9 11 7 10
Source: Barclays Capital
Note: Base Scenario assumes that national home prices fall 10% going forward and also assume some base level of
modifications. In all other scenarios we start with this base scenario and modify it as described above.
21 December 2009 60
Barclays Capital | U.S. Securitized Products Outlook 2010
CMBS
This should lead to a growing income gap, or pressure on borrowers to make debt
Tee Yong Chew payments given declining cash flows. The pace of defaults/modifications is set to
+1 212 412 2439 escalate in 2010, with cumulative defaults for 2005+ vintages doubling by the end
tee-yong.chew@barcap.com of the year to nearly 13%.
For loans that make it to balloon maturity, we still see pressure from the funding
gap, or the difference between existing mortgage debt and new mortgage
proceeds. This will lead to modifications, restructurings, and transfers of
ownership.
The expected level of stress will reveal structural gaps, testing legacy Pooling &
Service Agreements and the role of special servicers, leading to stubbornly high risk
premiums versus other asset classes. The restart of the new issue market, CMBS
2.0, provides a valuable opportunity to make improvements.
To begin the year, we recommend a positive stance on seniors and select AMs and a
negative view on subordinates. We also see attractive relative value trades across
CMBX. Toward H2 10, we see growing pressure as policy stimulus fades, rates rise,
and downgrades mount.
The title of last year’s outlook, Deleveraging, defaults and distress, will be largely applicable
again to CMBS and CRE markets in 2010 and the early part of this decade. The key difference
from last year is that we have a better gauge as to the depth of the economic downturn and
can see signs of recovery. Underlying commercial real estate fundamentals should start to
improve gradually with a lag, albeit from extremely weak levels. This should reduce tail risk
scenarios and support CMBS valuations at the top of the capital structure. However, a high
degree of stress is unavoidable for recent vintage CMBS, given the inherent lags in commercial
real estate and the magnitude of this cycle. This should put pressure on subordinate bond
pricing and cause considerable volatility across the capital structure in 2010.
This is a summarized version of the CMBS outlook. The full-length version will be published
shortly and available on Barclays Capital Live under keyword “cmbs.”
21 December 2009 61
Barclays Capital | U.S. Securitized Products Outlook 2010
Figure 1: YTD excess returns versus Treasuries, volatility Figure 2: Estimated loss-adjusted yields
US HY
Credit card 5y AAA
Auto 3y AAA
0
CMBS MBS Agencies ABS Corporate High
Yield
Note: Data YTD through December 18, 2009. Source: Barclays Capital Note: CMBS yields represent equal weighted average for 2005+ LCF dupers.
Estimated numbers. Source: Barclays Capital
depression-type scenario, which did not materialize. The removal of this outcome alone
was a significant positive for tail-risk-like securities 5 such as dupers.
In terms of our 2009 recommendations, we began the year with a neutral stance on the
basis, but we recommended buying recent vintage second- and third-pay AAA classes,
which we believed were structurally superior to last cash flow bonds but traded at much
wider spreads. This trade worked well, as S&P downgraded many recent vintage last cash
flow dupers, but left second- and third-pay classes largely immune. We turned overweight
on the overall sector on March 13 at the depths of the crisis and just before policy support
emerged in the form of PPIP/TALF. This was opportune, as the CMBS component of the US
Aggregate delivered +1,368bp of excess returns from March 13 to May 4. We failed to
appreciate the magnitude of the rebound fully, especially in AMs and AJs, and moved back
to a neutral stance in May, owing to growing concerns about credit performance, and
missed out on the late summer/early fall rally. We returned to an overweight with an up-in-
credit focus in early December, after CMBS lagged other sectors in November. Throughout
the year, we took advantage of several relative value opportunities across the capital
structure and different CMBX series; this will continue to be a focus in 2010.
5
By “tail-risk” type securities, we do not mean to imply that recent vintage dupers are immune to the probability of
loss; weaker LCF dupers take losses in our base case scenario. However, even in these cases, we expect them to be
minimal and more than offset by current pricing.
21 December 2009 62
Barclays Capital | U.S. Securitized Products Outlook 2010
Figure 3: Jobs outlook improving, should help demand for space Figure 4: But impact will be felt with considerable lags
3% 600 3%
200
2% 400 2%
100
1%
200 11 qt rs 9 qt rs 1%
0 0%
OFF 0 0%
-100 HOT -1%
RET -200 -1%
IND -2%
-200
APT -3% -400 -2%
YOY % Chg in Payrolls (rhs)
-300 -600 Off employment, qtrly chg in jobs (000s) -3%
-4%
PPR quarterly office rent growth RHS
-400 -5% -800 -4%
90 91 93 94 95 96 98 99 00 01 03 04 05 06 08 09 10 11 90 91 92 94 95 96 97 99 00 01 02 04 05 06 07 09
Note: Shaded area denotes recessionary period. Source: PPR, BLS Source: PPR, BLS
6
As measured by the sum of professional and business services and financial activities jobs, office using employment
appears set to turn positive in Q4 09 barring large revisions, with a net +104k jobs added in the two months ending
November 2009.
21 December 2009 63
Barclays Capital | U.S. Securitized Products Outlook 2010
shifting composition of retail sales, including a long-term migration toward online retailing.
The y/y change in non-store retailers, a proxy for internet spending, is up 8% through
November 2009, versus -5% for department stores. The non-store retail sales channel now
represents 8.6% of retail ex-auto sales, up from 8.1% a year ago. We believe this is a secular
phenomenon, which, along with a thriftier consumer, will prove a headwind for retail property.
For office, we are concerned about a secular decline in the size of the financial services
sector, which could have an outsized effect on major office markets in CMBS such as New
York City (NYC)7. Domestic financial sector debt peaked at 120% of GDP in Q1 09, up from
Long-term deleveraging in 81% in Q4 00 and 20% in Q4 90. Now, the trend is reversing as the financial sector
financial sector deleverages. In NYC, we find that financial activities jobs have declined 45k from the cyclical
peak of 474k in August 2007. The finance sector is critical to NYC office, given the multiplier
effect estimated to be at least 2:1; that is, two new related jobs for every one new job in the
securities industry.8 Despite recent signs of moderation, the NYC Independent Budget Office
expects more declines in financial activities jobs in NYC, hitting a trough at -59k through Q2
12, with more than two-thirds from the securities industry.9
7
NYC MSA office exposure represents 8% of the fixed rate CMBS universe, including 24.5% of the office sector.
8
Estimate from US Department of Commerce Regional Input-Output Modeling System.
9
New York City Independent Budget Office Fiscal Outlook, December 2009, A Cautiously Better Outlook: Fewer Job
Losses, Higher Tax Revenues.
10
Joint Center for Housing Studies, Harvard University, Household Projections in Retrospect and Prospect: Lessons
Learned and Applied to New 2005-2025 Projections, George S. Masnick and Eric S. Belsky, July 2009 W09-5.
21 December 2009 64
Barclays Capital | U.S. Securitized Products Outlook 2010
Construction and development A key distinction between bank CRE loan portfolios and CMBS is the prevalence of
loans are key risk construction and development loans (C&D). According to the FDIC, 27% of US bank CRE
loan exposure is in the form of C&D loans, or $492bn. Delinquencies on C&D loans are
running much higher than on more stabilized properties, given that the underlying collateral
for many such loans is not fully leased before completion. The rise in delinquencies has
occurred despite that fact that debt payments have benefited from low Libor rates, as the
majority of C&D loans (and bank CRE loans broadly) have floating-rate coupons. We expect
much higher losses on average for these loans than for more stabilized counterparts.
Volume of bank failures to rise Given the stress on C&D loans and overall lagged effect of CRE fundamentals, bank CRE
lending will remain under pressure, raising the need for resurgence in CMBS. The pace of
bank closures, which has totaled 140 through in 2009 through December 18, should
increase in 2010, which could provide opportunities for distressed investors. An improving
economic environment would certainly help offset bank losses, but if it came with a material
rise in short rates, this could be negated.
Index value 0%
200
-10%
175 -20%
150 -30%
75 -60% -55%
00 01 02 03 04 05 06 07 08 Moody's CPPI Est. "healthy" Est. "distressed"
Index property subset property subset
Office Retail Industrial Apartment
Source: Moody’s CPPI Source: Moody’s CPPI, Real Capital Analytics, Prof. David Geltner
21 December 2009 65
Barclays Capital | U.S. Securitized Products Outlook 2010
Expect private CMBS issuance to Whether or not TALF is extended, we do not expect this alone to have a material effect on
be <$20bn in 2010 spreads unless there were significant changes to each program. We do not expect major
new policy developments for CRE/CMBS in 2010 in the same magnitude as TALF. One
possible change would be a relaxation of the Foreign Investment in Real Property Tax Act
(FIRPTA), a 10% withholding tax on the sale of domestic property by foreigners, to spur
foreign investment.
We forecast only a gradual opening of the CMBS market, with 2010 private domestic
issuance to be less than $20bn, up from $2bn in 2009. We expect initial deals to be single
borrower, agented transactions with a slower rebound in conduit issuance.
Income gaps
Our biggest concern heading into 2010 is not the well-publicized loan maturities and
funding gap, but instead the income gap – that is, the growing pressure on legacy loans to
cover mortgage payments and avoid default. We start with a recap of the current
delinquency environment. With approximately 97% of December remittances reporting at
time of publication, 30+ day delinquencies have risen to 6.5% across the fixed rate universe,
or $43.5bn notional. We find another 3% that are specially serviced current. Many of these
loans are in significant distress, and involve borrowers attempting to negotiate some type of
Still early stage of the rise in
modification. The largest loan in this bucket is the $3bn Peter Cooper & Stuyvesant Town
delinquencies
loan, which transferred to special servicing in November.
We are still in the early stage of the rise in delinquencies and/or modifications. The primary
reason is the lagged effect of deteriorating CRE fundamentals, as highlighted above. With the
exception of hotel, we have seen only modest declines in net cash flow across our database of
CMBS loans. In office, the largest property type, average cash flows are down only 4% from
the peak, versus our peak-to-trough base forecast of -20 to -25%. From our perspective, the
bulk of delinquencies in 2009 has been more of a function of aggressive, pro forma
underwriting than from actual deterioration in cash flow performance, with the exception of
hotel. We estimate that $63bn of 2005+ vintage CMBS loans were pro forma. The percentage
of these loans in special servicing is 18%, versus only 7% for the non-pro forma loans. In
2010, we expect this to change.
Figure 7: Delinquent and special serviced loans rise Figure 8: Led by hotel and apartment, so far
2 2%
0 0%
1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 Office Retail Multifamily Hotel Industrial TOTAL
30+ day delinquent Specially serviced, current loans Dec-08 Jun-09 Dec-09
Note: Through December 2009, with 97% of December remittances reported; Note: Through December 2009, with 97% of December remittances reported;
fixed rate universe. Source: Barclays Capital fixed rate universe. Source: Barclays Capital
21 December 2009 66
Barclays Capital | U.S. Securitized Products Outlook 2010
Figure 9: Average change in net cash flow, fixed rate CMBS Figure 10: Estimate of current, but “high-risk” loans, 2005+
universe vintages
10% 0% 30
0% Total = $81bn or 18% of 2005+ vintage
-2%-2% -2%-2% 25 current loans
-10% -4% -3% -6%-2%
-8% Most recent DSCR <1.0X = $55bn
-20% 20 9 Most recent DSCR <1-1.1X = $26bn
-30% -24%
5
-40% 15 7
-50% 2
10
-60% -53%
16 14
Office
Multifamily
Other
Retail
Hotel
Industrial
5 11 10
2
1
2 3
0
Office Retail Multifamily Hotel Industrial Other
From June 2007 through Sep 2009 YTD through Sep 2009 <1.0X 1-1.1X
Note: Fixed rate universe. Source: Barclays Capital Note: Fixed rate universe, as of Nov. 2009. Source: Barclays Capital
Recent data supports our As cash flow declines materialize, many recent vintage loans that are current will face
expectation for a rise in pace of pressure. Across the 2005+ vintage universe of $454bn current loans, we find that $81bn, or
credit deterioration 18%, have a most recent DSCR <1.1X at the A-note level, including $55bn <1.0. Unlike prior
cycles in CMBS, larger loans show the most significant stress, and will be more difficult to
modify/restructure given the presence of numerous parties involved. Given the decline in
property values highlighted previously, many of these loans are underwater, which suggests
that there is limited incentive to fund a prolonged shortfall. We believe a much higher
percentage of such borrowers will default in this cycle than in prior cycles and remain
sceptical of legacy CMBS default models based on the limited data history covering more
benign cycles. Recent data tend to support this view of an increase in the pace of
delinquencies; for example, 30+ day delinquencies surged +69bp in December for 2005+
vintage loans, versus the prior 3-month average of +41bp.
50
40
30
20
10
0
Jan-10 Jan-12 Jan-14 Jan-16 Jan-18 Jan-20 Jan-22 Jan-24
Note: Cumulative default figures expressed as % of cutoff balance. This does not include loans that extend but do not
take a loss. If a loan extends but eventually does take a loss, the default date represents when the loan was extended.
Source: Barclays Capital
21 December 2009 67
Barclays Capital | U.S. Securitized Products Outlook 2010
For these reasons, we expect defaults to rise sharply. For 2005+ vintages, we forecast defaults,
including substantially modified loans, to double to nearly 13% by the end of 2010. We
caution against the usefulness of traditional delinquency measures, given our expectation of
modifications that return delinquent loans to current, as well as a pickup in loans exiting the
delinquency pool from liquidations. On the former, we will begin tracking modified-current
loans next year, defined as any delinquent loan that had a modification to principal or interest
payments. Liquidation timelines will likely remain extended, but the rise in delinquencies should
quickly lead to an escalation of interest shortfalls. On average, we expect these to rise into
investment grade classes for 2005+ vintage deals next year, with select AJs taking shortfalls by
the end of the year.
Funding gaps
For loans that make it to balloon maturity, we still see pressure from the funding gap, or the
difference between existing mortgage debt and new mortgage proceeds. This funding gap
is a function of both declining cash flows and tighter underwriting standards. As shown in
Figure 12, we estimate that nearly $1.5trn of CRE maturities will come due by 2012. We
make assumptions about the average life of the non-CMBS component and assume that
bank C&D loan maturities are equally dispersed over the next two years. The CMBS
exposure is more back-ended, with the exception of floating rate CMBS, where maturities
peak in 2011.11
$ Bln Other
GSE's & Fed Related mtg Pools
600 Insurance Companies
Commercial banks, savings inst
500 Floating Rate CMBS
Fixed rate CMBS
400
300
200
100
0
2010 2011 2012 2013 2014 2015 2016 2017
Maturity
Note: With the exception of CMBS maturities, the rest are estimated figures. Source: Barclays Capital
Estimate the funding gap across Unfortunately, we do not have detailed statistics across all CRE loans; however, we can
fixed rate CMBS is over $120bn closely track the behavior of loans in CMBS deals. YTD through November 2009, we find
that $9.4bn of maturing CMBS loans have paid off net of defeased loans, or 56%
(Figure 13). If we look closer, we find a big distinction between vintage, with nearly 70% of
pre-2002 vintage loans paid off versus less than 40% for 2002+ vintage loans. We do not
yet see any signs of a pickup in lending activity, as the trailing 3-month average has been
roughly flat; this will be an important metric to follow in 2010.
11
Floating-rate loan maturities based on maximum extension terms.
21 December 2009 68
Barclays Capital | U.S. Securitized Products Outlook 2010
Figure 13: 2009 CMBS maturing loans struggled Figure 14: Debt yield distribution of maturing loans
Note: Fixed rate universe. Source: Barclays Capital Note: Fixed rate universe. Source: Barclays Capital
We attempt to estimate the size of the funding gap across the fixed rate CMBS universe. We
focus on NOI debt yield, or the trailing 12-month NOI estimate divided by the A-note
current loan balance. Historically, attractive exit debt yields have been about 12%. The three
new CMBS 2.0 deals were all 15%+, reflecting today’s tighter underwriting standards.
Across the fixed rate universe, we estimate that 70% of loans have a current NOI debt yield
<12%, including 85% of loans scheduled to mature in 2017 (Figure 14). To measure the size
of the funding gap in fixed rate CMBS, we solve for the required balance to produce a 12%
debt yield based on most recent NOI, and subtract the actual loan balance. Across the fixed
rate universe, the funding gap is $123bn, or 20% by current balance of all loans. In 2010, we
see a $6bn shortfall for the $32bn of maturing fixed rate loans; in 2017, this rises to $35bn.
We suspect that the actual gap will be much lower, given our view on term defaults and
modifications as highlighted earlier.
Quality of maturing loans The optimistic case would be that there is plenty of time for fundamentals to recover and
declines each year cash flows to improve. However, this can be a dangerous assumption. Many of the recent
vintage loans are partial or full-term interest only (IO) loans. We suspect that many partial
IO loans from 2005+ vintage will be converted to full-term IOs, so the loans will not benefit
from deleveraging over time. Also, the debt yield numbers are still trending lower, given the
lagged effects on cash flows of most property types as highlighted earlier. The persistence
of this funding gap will cast a cloud over CRE markets over the secular horizon.
Group A: loans that are covering existing debt service but cannot (or are unwilling) to
find new financing at current market rates. These will lead to straightforward
extensions; the only decision is as to how long an extension and what will it cost.
Group B: loans that are term default candidates, where current cash flow is insufficient
to cover debt service. These modifications will be much more complex.
Group A modifications will likely be the least controversial, as it is better to work with these
borrowers than to foreclose and liquidate. The key variables will be the length of the
extension and the magnitude of any step up in coupon or principal pay-down required. The
21 December 2009 69
Barclays Capital | U.S. Securitized Products Outlook 2010
recent GGP proposal provides a blueprint that we expect to apply to most other
modifications. It confirmed our view that extensions will need to be much longer than most
expect; the average extension for the GGP loans in the proposal was more than 4.8 years.
We were disappointed with minimal step-up in amortization, as the extension terms were
well below current market rates to the benefit of unsecured debt and equity investors at the
expense of the CMBS trusts.
Finally, for both groups, a key concern for us in 2010 is how accurately these modifications
are reported. As we highlighted on December 11, 2009, we have seen only sporadic
completion of the loan modification report as required by the CMSA investor reporting
package so far.
Structural gaps
Legacy deal structures put to the test
Another major theme heading into 2010 is structural gaps. Simply put, legacy CMBS deal
structures were not designed to handle the magnitude of stress that we expect in 2010.
The uncertainty from these structures impedes fundamental analysis, and will cause
investors to continue to demand higher risk premiums from legacy CMBS than versus other
asset classes.
Conflicting interests in A major concern exists regarding incentives. Special servicers are responsible for working
legacy structures out all troubled loans. Usually, they also own the first loss piece of the deal. The idea was to
align the incentives of the special servicer with the rest of the trust. However, for most
recent vintage deals, this first loss piece is essentially a credit IO strip; that is, there is very
limited chance for any principal recovery. As long as the first loss piece has not been written
down from losses, the B-piece buyer still remains the controlling class and has the ability to
appoint the special servicer. In such a scenario, servicers have an incentive to extend the
length of that IO strip and maintain control as long as possible, while adhering to the rules
of the Pooling & Service Agreement (PSA) and maximize net present value. The discount
rate is typically below loss/extension-adjusted yields on longer-dated AAA bonds, pitting
AAA buyers versus special servicers in an adversarial relationship.
Test of non-recourse In 2009, there was also a challenge to the non-recourse nature of CMBS. GGP’s corporate
bankruptcy in April 2009 was not a surprise. However, the fact that 79 special purpose entities
associated with more than $9bn of CMBS loans followed them was a surprise. The filing of the
SPEs came despite adequate performance across most of these loans, with DSCR well above
1.0x. This put into question the role of the independent directors. The worst case outcome –
substantive consolidation – looks to have been avoided. However, the SPE bankruptcy filing was
a clear negative for CMBS and is something which will need to be priced in to future deals.
21 December 2009 70
Barclays Capital | U.S. Securitized Products Outlook 2010
Initial deals contain structural The DDR and Inland Western deals appear most similar in terms of features. Both allow for the
improvements replacement of the special servicer by someone other than the majority holder of the most
subordinate class, unlike legacy deals. The trigger can be due to either losses or appraisals.
Both deals also have new language about the discount rate used in the special servicer NPV
decision engine. In each case, they attempt to incorporate an estimate of the market rate for
similar debt. We believe this is a significant improvement over the legacy CMBS deals. Also,
the fact that investors felt comfortable “paying up” for these features means they will likely
become more common.
Relative value
We begin the year positive on We start the year positive on the basis, with an up-in-credit bias and a preference for recent
seniors, negative on vintage last cash flow AAAs. We also see value in select AMs, but remain firmly negative on
subordinates AJs and lower subordinate bond classes. Swelling delinquency pipelines, rising interest
shortfalls, and a move closer to eventual write-downs should weigh on subordinate bond
pricing in 2010. We expect to be active with our basis views, as we forecast considerable
volatility throughout the year. We also recommend adding short positions in CMBX
subordinate classes; our favorite outright short position in CMBX remains A.3.
We begin with a review of our CMBS loss expectations. Our base case cumulative loss
expectation for CMBX.4, which represents 2007 vintage collateral, is 16%; in our stress
case, this rises to 26%. Slightly seasoned vintages fare much better; for example, our base
case loss expectation for CMBX.1 (2005 vintage proxy) is only 8%. As highlighted earlier,
Figure 15: Base case deal loss, CMBX Series Figure 16: Stress case deal loss, CMBX Series
Cumulative Cumulative
40 40
deal loss deal loss
30 30 26
24
22
20 16 20 17
15 14
13
10
8 10
10
- -
Series 1 Series 2 Series 3 Series 4 Series 5 Series 1 Series 2 Series 3 Series 4 Series 5
21 December 2009 71
Barclays Capital | U.S. Securitized Products Outlook 2010
recent economic data have caused us to reduce our weighting on our stress case outcomes,
which helps more tail-risk securities such as dupers. Our loss expectations involve a hybrid
approach, with the assistance of our surveillance analysts manually evaluating larger, more
complex loans and a statistical model for the smaller loans based on NOI vectors by region
and property type.
Extension risk is a key concern Beyond our loss expectations, a key determinant of valuation is the timing of losses and
principal recovery. We do not expect a wave of liquidations in 2010, but to get closer to the
expected wave in 2011 and 2012. Our typical liquidation timeline for a defaulted loan is 36
months. We expect considerable extensions on maturing loans, as highlighted earlier. These
factors would cause recent vintage bonds to extend dramatically. We expect this to extend
beyond second- or third-pay classes; for 2007 vintage last cash flow dupers, our average
bond life is 8.9 years, versus 7.2 years at 0 CDR. For those that express long-term basis
trades (for example, long 2007 vintage dupers versus swaps), we strongly recommend
hedging to a longer, extension-adjusted model duration instead of 0 CDR duration.
Favor LCF dupers over second- Combining our view on losses and extensions, we calculate loss/extension adjusted yields
and third-pay classes across recent vintage CMBS (Figures 18 and 19). We see outright and relative value in
recent vintage last cash flow bonds, with base case yields of 5.6-9.2%. We believe the
excess spread pickup versus other sectors such as ABS and corporates is required to offset
exceptional volatility in CMBS. Within CMBS, we see better value in last cash flow dupers
from 2006 and 2007, versus second-pay classes, unlike last year. In recent months, there
has been a pickup in crossover interest in CMBS from investment grade corporate and high
yield investors. We expect this to continue in 2010, providing support for the sector.
For AMs and AJs, our views are much more deal specific, as there is tremendous variation.
We like select AMs but remain largely negative on 2006+ AJs, which stand to take significant
writedowns in our base case scenarios. Also, we continue to be sceptical about the amount
of demand from PPIP managers at current pricing.
21 December 2009 72
Barclays Capital | U.S. Securitized Products Outlook 2010
2005+ vintage last cash flow dupers by count, versus only 3% for Moody’s and 0% for Fitch
and Realpoint (Figure 20). We see a similar trend in AMs and AJs. Given our view on
delinquencies, we expect this gap to narrow in 2010, with Moody’s and Fitch taking a more
aggressive stance. There are some signs that this is already occurring: Moody’s downgraded
four recent vintage LCF dupers. Also, Fitch recently placed $20.6bn of bonds from 33
floating-rate deals on negative ratings watch, including $14bn rated AAA.
Insurance companies could be A more aggressive stance could cause selling pressure in 2010, especially for AMs and AJs.
sellers of AMs and AJs in 2010 The main seller could be insurance companies, which face an increase in regulatory capital
requirements for bonds downgraded to BBB or lower. Rating determinations are based on the
“lower of two or middle of three” methodology, so the effect of S&P’s downgrades has been
minimal in isolation. According to Moody’s,12 insurance companies held approximately
$200bn notional of CMBS at year-end 2008. We estimate that approximately $37bn of the
$200bn was 2005+ AMs and AJs. We assume that these numbers held constant in 2009,
which is not unreasonable, given that we did not see a wave of insurance company selling. If
Moody’s or Fitch were to downgrade to the same extent as S&P, up to $20bn of the originally
rated AAA bonds (that is, dupers, AMs and AJs) held by insurance companies could fall to BBB
or below by two or more agencies. This would lead to higher capital requirements and could
provide a source of supply for PPIP managers at more attractive prices.
Figure 18: Loss-adjusted yields, senior AAAs Figure 19: Loss-adjusted yields, AMs & AJs
12
U.S. Life Insurers’ Commercial Mortgage Exposure & Losses are Manageable, Moody’s Global Insurance Special
Comment, December 2009.
21 December 2009 73
Barclays Capital | U.S. Securitized Products Outlook 2010
Figure 20: 2005+ vintage LCF dupers, % D/G by count Figure 21: 2005+ vintage AMs/AJs, % D/G by count
80% 80%
60% 4%
100% 60%
95%
40%
40%
51% 46%
20%
3% 20%
0%
AAA IG Non- AAA IG Non- AAA IG Non- 0%
IG IG IG S&P Moodys Fitch S&P Moodys Fitch
real estate values if not offset by a commensurate improve in growth prospects. The second
involves our view on extensions. If CMBS spreads remain unchanged, higher rates would
push most longer-dated AAA bonds further into discount territory, putting downward
pressure on loss/extension-adjusted spreads.
Figure 22: Buy AJ.2, sell AJ.5 Figure 23: Buy A.4, sell A.3
50
30 4
40
25 2
30
20 0
20
10 15 -2
0 10 -4
Jan-09 Feb-09 Apr-09 Jun-09 Jul-09 Sep-09 Nov-09 Jan-09 Feb-09 Apr-09 Jun-09 Jul-09 Sep-09 Nov-09
Source: Barclays Capital Source: Barclays Capital
21 December 2009 74
Barclays Capital | U.S. Securitized Products Outlook 2010
CONSUMER ABS
The biggest risks to the sector in 2010 are regulatory and legislative. Given the
mantra of consumer protection in Washington, there is a significant risk of over-
regulation occurring, with negative ramifications for consumer ABS performance.
Additionally, the drive by Congress and the FDIC to place new requirements on ABS
issuers could hamper securitization if not done thoughtfully.
The improving economy in 2010 should help stabilize consumer ABS collateral
performance, albeit it at still weak levels. We expect credit card charge-offs to peak
at 11.0-11.5% in Q1 10, but trend down to 9.0-9.5% by year-end. Retail auto
delinquencies and losses will likely remain range bound, but at levels lower than the
peaks of 2009.
We expect the market to handle the expiration of TALF in stride, with 2010 issuance
volume of $120-135bn, flat to slightly lower than 2009 and well below the peak of
2005-07. The relatively low supply, combined with large amounts of investor cash
on the sidelines and the re-emergence of the asset class as a safe haven, should keep
the tightening trend intact. We believe mezzanine and subordinate credit card ABS
offer the best relative value, but we also like recent vintage retail auto lease ABS.
Regulatory risk
New requirements for The foremost regulatory risk we see relates to the FDIC’s deliberations on modyfing its
securitizers likely as part of the securitization safe harbor; we believe this will have wide-ranging implications for all
FDIC safe harbor extension securitizations, not just consumer ABS. At its December board meeting, the FDIC approved an
Advance Notice of Proposed Rulemaking (ANPR) relating to its treatment of securitizations
after the interim extension of the Securitization Rule’s safe harbor expires on March 31, 2010
(see Consumer ABS Strategy Update: FDIC Acts on Securitization Rule, Advanta Corp. Files for
Bankruptcy Protection, November 13, 2009 for a detailed discussion of the extension). The
FDIC will accept comments on the ANPR for 45 days from date of publication in the Federal
Register, after which it plans to publish a full Notice of Proposed Rulemaking (NPR). The fact
that it issued an ANPR, as opposed to an NPR, is noteworthy, as NPRs are typically not
significantly revised once published.
21 December 2009 75
Barclays Capital | U.S. Securitized Products Outlook 2010
Risk retention, additional The FDIC provides an example of language that it could use as a template for a final rule in
disclosures, and enhanced deal the appendix of the ANPR. However, throughout the document, the FDIC goes to great
documentation likely for pains to point out that the ANPR is not a formal rule proposal, but rather just an example
consumer ABS designed to generate discussion (the same qualification came up several times during the
actual board meeting, as well). Though the ANPR is truly a request for comment, and the
appendix represents only sample language, it is nevertheless apparent that the FDIC has
serious reservations about future securitizations (particularly mortgage-related). For non-
mortgage securitizations, the sample language suggests risk retention (5%), additional data
disclosures, and enhanced deal documentation. The shape of the final rule will no doubt
have important ramifications for the future of the consumer ABS market; however, based on
the sample language, other than risk retention, we do not see anything too onerous for
consumer ABS. With the possible exception of credit card securitizations, issuers of which
already retain risk through the seller’s interest, a 5% retention requirement will likely make
securitization economically difficult for most issuers if enacted across the board.
Requirements likely to apply to The good news is that the FDIC appears willing to accept comments and structure a rule
all securitizers, not just those ensuring that securitization remains an important tool. However, it is moving ahead quickly
that are FDIC regulated on this initiative, quicker than Congress, in fact. Clearly, the rule would apply to bank issuers
only, but the FDIC is working closely with other regulators in crafting it, so we expect
significant coordination; indeed, during the Board meeting, members indicated a strong
desire that all securitizers be subject to the same rules and compete on a level playing field.
The comment period for the ANPR runs into early/mid-February, and we figure the earliest
the FDIC can get an NPR out is at the April Board meeting. Given the typical 45-day
comment period, we expect new regulations not to be implemented until July 2010 at least.
Thus, in our view, it is imperative for the FDIC to extend its interim safe harbor provision.
Legislative risk
In addition to new requirements and stipulations that may come about through regulatory (or
legislative) fiat, we see four major legislative initiatives on the calendar that are likely to affect
consumer ABS in 2010: creation of the Consumer Financial Protection Agency, interchange
legislation, bankruptcy cramdown legislation, and implementation of the CARD Act.
Consumer Financial Protection On December 2, 2009, Representative Barney Frank, Chairman of the House Financial
Agency awaits Senate action Services Committee, introduced H.R. 4173, the Wall Street Reform and Consumer Protection
Act of 2009, which was passed by the full House on December 11. The legislation sets forth
a sweeping set of measures to overhaul regulations in the financial sector, including a risk
retention requirement for securitizers. Also incorporated into the bill is the Consumer
Financial Protection Agency, an independent federal agency, originally introduced in H.R.
3126, with supervisory, examination, and enforcement authority over consumer financial
products or services (eg, mortgages, credit cards, payday loans, terms on savings accounts,
et al). As envisioned in H.R. 4137, the CFPA will be headed by a single director, appointed by
the president, with the advice and consent of the US Senate, for a term of five years. The
director will have full executive authority to issue regulations and take other actions to carry
out purposes of the act. The director will be advised by the Consumer Protection Oversight
Board, consisting of 12 members: seven state and federal banking regulators and five
additional members. The board will not have any executive authority.
The CFPA is still in the formative stages, as the Senate has yet to act. However, given the
strong pro-consumer/anti-banking sentiment on Capitol Hill, it seems likely that some
version of new financial services regulations will pass in 2010. Whether or not a single
21 December 2009 76
Barclays Capital | U.S. Securitized Products Outlook 2010
super-regulator for financial services is included, investors need to consider the changing
regulatory landscape that is likely to develop.
Interchange legislation Interchange legislation has been on the docket since the summer, with Congress having
asked the General Accounting Office (GAO) to study the issue earlier in the year. On
November 19, the GAO released its results. In our view, the report was relatively neutral and
inconclusive, emphasizing the complexity of the issue and not really pointing a clear way
forward. It indicates that interchange rates have increased over time in part due to card
networks trying to attract more issuers as customers, but it also recognizes that cardholders
and merchants have benefitted from greater plastic use. The study presented mixed
evidence about whether lower interchange fees would ultimately benefit the consumer.
Merchants may be able to use savings from lower interchange to reduce prices for the
goods and services they sell. However, the study noted that issuers reported relying
substantially on interchange to fund their credit card operations, which suggests that they
would need to raise interest rates and/or other fees in response to regulation. The study
explored four options to regulate merchant card acceptance costs:
Restricting card networks from imposing rules on merchants (eg, preventing merchants
from imposing a surcharge for using credit cards)
The report suggested that either setting/limiting interchange fees or restricting card
networks from imposing rules on merchants would be easiest to implement, but recognized
that it will be difficult to determine a fair rate and that cardholders may be affected by
merchants' ability to impose surcharges on cards. The report highlighted a 2003 initiative by
the Reserve Bank of Australia, which required Visa and MasterCard to reduce interchange
fees from a 0.95% weighted average to 0.5% and allowed merchants to impose surcharges
for credit cards. In response, banks predictably increased other fees and lowered rewards.
We view any legislative action to curb interchange as a negative for credit card ABS trust
yields, but expect issuers to compensate for the reduction through increasing other fees
and/or reducing costs (eg, scaling back or eliminating reward/loyalty programs). However,
we believe interchange legislation remains unlikely because of the crowded legislative
calendar and the fact that the study does not represent interchange as clearly a consumer
issue. That said, we also believe that because the report offers support for both sides, the
issue will not go away and could continue to act as an overhang in 2010.
Bankruptcy cramdown could The on-again, off-again legislative initiative of the past year has been bankruptcy
reappear in 2010 cramdown. Originally used by Congress as a stick to push mortgage servicers to increase
loan modification activities early in 2009, the specter of bankruptcy cramdown reappeared
in December with an amendment offered to H.R. 4137. The current incarnation is identical
to one passed by the House in March but subsequently defeated in the Senate. The House
rejected the amendment when considering (and ultimately passing) H.R. 4137 on
December 11. Nevertheless, we do not believe the issue is dead. Evidence has been
mounting of the slow pace and effectiveness of mortgage loan modifications. To the extent
this trend continues, we believe cramdown legislation could be revived in 2010, with
Congress arguing that it has no choice but to allow bankruptcy judges to change mortgage
terms, given the industry’s inability to help struggling homeowners. If such legislation were
enacted, the spillover effects would be most felt in credit card securitizations, as
21 December 2009 77
Barclays Capital | U.S. Securitized Products Outlook 2010
homeowners filing for bankruptcy to save their houses would also likely seek to discharge
their unsecured debt as well. We estimate that cramdown legislation could increase credit
card charge-offs 200-400bp.
The CARD Act The CARD Act, signed into law by President Obama on May 22, will become fully effective
on February 22, 2010. The most onerous provisions for consumer ABS have already been
implemented (namely, limits on risk-based pricing and finance charge increases).
Nevertheless, compliance with the new law will likely cause hardship. We have already seen
the initial effects, which were largely as advertised: less credit availability, at a higher cost.
Full implementation would only continue this trend. As such, through 2010 we expect credit
card issuers to remain selective in granting credit, to tighten availability at the lower end,
and to charge more for the credit they do extend.
TALF for consumer ABS Despite (or, more accurately, because of) the success of the program, we expect the Fed to
likely to expire allow the consumer-related portions of TALF to expire as scheduled on March 31, 2010.
18 18
16 16
14 14
12 12
10 10
8 8
6 6
4 4
2 2
0 0
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
Auto Credit Card Student loan Floorplan Auto Credit Card Student loan Floorplan
Equipment Mtg Svcg Adv Insurance Equipment Mtg Svcg Adv Insurance
Source: IFR Markets, Barclays Capital Source: IFR Markets, Barclays Capital
21 December 2009 78
Barclays Capital | U.S. Securitized Products Outlook 2010
Exceptions might include some of the more off-the-run asset classes, but the decision will
come down to whether the government wants to continue to subsidize these sectors.
Nevertheless, the Fed has appeared concerned about its risk position from TALF for some
time now (eg, adding risk assessments for consumer ABS), and we believe it would be hard-
pressed to find the “exigent and unusual” circumstances in today’s market necessary to
justify continuing the program statutorily.
2010 issuance volume will likely This begs the question: how will the consumer ABS market fare in a post-TALF world? We
be flat to slightly lower do not view the expiration as a major destabilizing event. However, it could have minor
implications for volatility near the end of Q1 10. New issue volume in 2009 for the primary
consumer ABS classes (ie, auto, credit card, equipment, and student loans) was $128.1bn, a
2.4% increase over 2008’s $125.1bn (which followed a 44% decline from $216.7bn in
2007) and was distributed as shown in Figure 3. Given the expiration of TALF, the various
regulatory and legislative headwinds facing consumer ABS issuance (changed accounting
regime, loss of regulatory capital relief, uncertainty about risk retention requirements), and
the incipient economic recovery, we expect aggregate new issue consumer ABS to be flat to
slightly lower in 2010. Specifically, we expect new issue volume of $65-70bn for autos, $35-
40bn for credit cards, $5-8bn for equipment, and $15-20bn for student loans.
Card
42,671
33.3%
Auto
58,198
45.4%
Student Loan
Equipment 18,062
9,216 14.1%
7.2%
21 December 2009 79
Barclays Capital | U.S. Securitized Products Outlook 2010
Figure 4: Consumer ABS spreads: Current versus 2y average pre-Lehman bankruptcy +/- one standard deviation
750
650
550
450
350
250
150
50
-50 Autos Detroit 3 Cards-Fix Cards-Flt Autos Detroit 3 Cards-Fix Cards-Flt Autos Detroit 3 Cards-Fix Cards-Flt
Mezzanine and subordinate We quantify the potential for additional spread tightening in Figure 4, which plots current
credit card ABS offer the most consumer ABS spreads for prime retail auto, Detroit three auto, and fixed- and floating-rate
attractive value credit card ABS, as well as the +/- one standard deviation range for the average spread over
the 2y period prior to Lehman’s bankruptcy. The graph suggests that mezzanine and
subordinate credit card ABS offer the greatest spread tightening potential. Our analysis
assumes that spreads will revert to slightly higher-than-historical norms; we believe it is more
likely that nominal spreads will settle into a new range above the pre-crisis level, rather than
return to historical long-run averages. Without being so presumptuous as to think we can
determine the “new normal” consumer ABS spread level, our analysis leads us to believe that
fixed and floating rate credit card ABS offer the best relative value heading into 2010. For a
safe cash alternative, we like senior short-dated retail auto and credit card ABS.
We also recommend less In the search for incremental yield within the non-mortgage ABS sectors, we also
common asset classes such as recommend investors turn to more off-the-run asset classes, including dealer floorplan,
recent vintage retail auto lease recent vintage retail auto lease, and rental fleet securitizations. Such asset classes offer
attractive yield pickup relative to generic consumer ABS, with a minimal increase in risk.
Dealer floorplan issues offer unleveraged yields of 2.5-3.0%, and their structures have been
tested with the GM and Chrysler bankruptcy filings. We recognize that the government was
heavily involved in steering both companies through the bankruptcy filing process smoothly
and that such benevolent intervention may not be forthcoming again. Nevertheless, we
21 December 2009 80
Barclays Capital | U.S. Securitized Products Outlook 2010
believe the dealer floorplan structure is sound enough for the senior notes to withstand a
messier bankruptcy process than GM and Chrysler endured.
We also like recent vintage retail auto lease transactions, which offer a yield pickup of 30-
40bp over retail auto loan ABS. The used car market suffered significant declines earlier in
the year, resulting in decreasing residual values on consumer auto leases. Retail lease ABS in
2009 were structured with these lower residual value leases and were required to have
higher credit enhancement by more conservative rating agencies. We believe this
combination increases the credit quality of 2009 lease deals, which continue to trade cheap
relative to auto loan ABS. Recent rental fleet securitizations have also come with very high
initial enhancement (upwards of 50% on HERTZ 2009-2) and continue to trade with a
significant spread pickup of 175-200bp over retail auto and credit card ABS.
Figure 5: Auto ABS issuance rebounded in 2009, but remains below historical levels
100
80
60
40
20
0
2001 2002 2003 2004 2005 2006 2007 2008 2009
Source: IFR Markets
We expect a 15-20% increase in Aside from potential regulatory/legislative headwinds impeding securitization (predominantly
auto ABS issuance the 5% risk retention requirement) discussed in Themes for 2010 above, new vehicle sales and
the expiration of TALF might also have an effect on 2010 new issue volume. The average of
five forecasts (CSM Worldwide Inc., J.D. Power and Associates, Edmunds.com, IHS Global
Insight Inc., and Center for Automotive Research) for 2010 new vehicle sales is 11.7mn units,
compared with expected 2009 sales of 10.3mn. Higher sales bode well for 2010 retail auto
loan and lease ABS issuance. In our view, the expiration of TALF should not adversely affect
such issuance, as most recent transactions (even those that were TALF eligible) were not
21 December 2009 81
Barclays Capital | U.S. Securitized Products Outlook 2010
financed through TALF. The fact that several deals were brought to market outside of TALF in
November also gives us comfort. Dealer floorplan is likely to be a trickier proposition; the vast
majority of the 2009 volume was likely financed through TALF. As such, we are not convinced
that the broader market for such deals can stand on its own without the support of TALF.
Foreign captive and independent finance companies will likely have the ability to tap the
market. However, we believe that in the absence of TALF leverage, others will have to offer
higher spreads to attract investors. In the aggregate, we expect $65-70bn in new issuance in
2010, an increase of 15-20% over 2009, with the lion’s share coming from prime retail.
Collateral performance likely to Retail auto collateral performance continued to deteriorate in 2009, consistent with our
improve slightly expectations (Figures 6 and 8). Early in the year, delinquencies and annualized net losses
reached all-time highs (0.88% and 2.27% for prime and 5.3% and 12.9% for non-prime,
respectively). However, during the spring and early summer seasonal improvement,
performance retraced much of the deterioration. While subprime performance has exhibited
seasonal weakness through the fall, prime collateral has not. Additional signs of stabilization
are evident in seasonally adjusted performance (Figures 7 and 9). On a seasonally adjusted
basis, late payments and annualized net losses have been declining. This suggests the
performance improvements are due to more than just seasonal factors and gives us
comfort with our 2010 performance forecast: we expect prime delinquencies to peak at
0.8% in February/March with a trough of 0.5% in mid-summer, and losses to have a high
and a low of 1.9% and 1.6% over the same period. For non-prime, we expect delinquencies
to hit 5.0% in early 2010 and drop to 3.5% in mid-summer, with annualized net losses
posting a high and low of 11.5% and 8.0%, respectively.
Spreads tightened throughout Since the announcement of TALF in November 2008 and the release of program details in
2009… December 2008, auto ABS spreads have tightened dramatically. Figure 10 details the steady
march tighter of prime retail auto ABS in 2009. Similar spread performance was observed in
subprime retail auto loan, retail lease, and dealer floorplan spreads. Tightening in retail auto
ABS was largely due to investors regaining comfort in the credit performance and structural
integrity of the asset classes, while improved spread performance in dealer floorplan ABS
can largely be credited to the smooth bankruptcy filings of GM and Chrysler.
…and we expect the trend to There are far fewer clouds on the horizon for the auto ABS market heading into 2010 than
continue, but at a slower pace there were at the start of 2009. We see room for additional spread improvement as
investors continue to gravitate to the asset class as a safe haven for cash. With our
expectation of increased issuance, lack of supply should not be a technical factor driving
spreads tighter, unlike other consumer asset classes. At the same time, we do not expect
the incremental volume to come anywhere near satiating investor demand for high-quality,
short-term, fixed-rate paper.
21 December 2009 82
Barclays Capital | U.S. Securitized Products Outlook 2010
Figure 6: Prime retail auto ABS collateral performance Figure 7: Seasonally adjusted prime performance
2.5% Ann net losses 60+ delinquencies 2.5% SA Ann net losses SA 60+ delinquencies
2.0% 2.0%
1.5% 1.5%
1.0% 1.0%
0.5% 0.5%
0.0% 0.0%
Nov-04 Nov-05 Nov-06 Nov-07 Nov-08 Nov-09 Nov-04 Nov-05 Nov-06 Nov-07 Nov-08 Nov-09
Source: Intex, Barclays Capital Source: Intex, Barclays Capital
Figure 8: Non-prime retail auto ABS collateral performance Figure 9: Seasonally adjusted non-prime performance
14.0% Ann net losses 60+ delinquencies 14.0% SA Ann net losses SA 60+ delinquencies
12.0% 12.0%
10.0% 10.0%
8.0% 8.0%
6.0% 6.0%
4.0% 4.0%
2.0% 2.0%
0.0% 0.0%
Nov-04 Nov-05 Nov-06 Nov-07 Nov-08 Nov-09 Nov-04 Nov-05 Nov-06 Nov-07 Nov-08 Nov-09
Source: Intex, Barclays Capital Source: Intex, Barclays Capital
Figure 10: Prime retail auto ABS spreads tightened steadily through 2009
2100
3yr AAA 3yr A 3yr BBB
1800
1500
1200
900
600
300
0
Jan-07 May-07 Sep-07 Jan-08 May-08 Sep-08 Jan-09 May-09 Sep-09
Source: Barclays Capital
21 December 2009 83
Barclays Capital | U.S. Securitized Products Outlook 2010
Figure 11: Credit card ABS issuance declined on FAS 166/167, FDIC uncertainty
Issuance volume likely to be Scheduled maturities of credit card ABS in 2010 total $92.0bn, compared with $64.9bn in
down despite significant 2009 and $67.0bn in 2008, but appear relatively well spread out throughout the year (Figure
maturities 12). Given the uncertain regulatory and legislative framework, and the strong pro-consumer
sentiment in Washington, we believe credit card issuance volumes will remain under
pressure in 2010. In addition, loss of regulatory capital relief (which already occurred for
issuers supporting transactions) will apply further downward pressure. Thus, despite the
significant amount of maturities, we expect only $35-40bn of new credit card ABS, the bulk
of which will likely be from bank card issuers, with the remaining maturities financed some
other way. In addition, a whopping 90% of maturities are senior class A notes, so there will
likely be less pressure to issue subordinate classes (publicly or privately) to maintain
required enhancement levels.
21 December 2009 84
Barclays Capital | U.S. Securitized Products Outlook 2010
14
12
10
-
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
Source: Bloomberg, Barclays Capital
Collateral performance likely to Bankruptcy filings and unemployment continue to be the main drivers of credit card charge-
stabilize as the economy offs. Our regressions of changes in non-business bankruptcy filings and the unemployment
improves rate against changes in credit card charge-offs result in adjusted r-squareds of 74% and
70%, respectively. Our analysis indicated that for an increase in bankruptcy filings of 10,000,
we would expect credit card charge-offs to increase about 33bp. Similarly, a 100bp rise in
the unemployment rate would result in an expected 99bp increase in charge-offs (Figure
13). The severe run-up in unemployment over the past 18 months has led this variable to
have more significance than bankruptcy filings recently. Given the larger-than-historical y/y
increases in the unemployment rate this year, we explored the unemployment rate/charge-
off relationship at higher y/y changes (Figure 14). While the data set is more limited, we
found that at these higher unemployment rates, the same 100bp increase resulted in
approximately 125-130bp of additional charge-offs. However, based on our linear
regression, and given our economists’ forecast for quarterly unemployment in 2010, we
expect credit card-charge-offs to plateau at 11.0-11.5% and gradually trend down to 9.0-
9.5% by the end of the year.
Figure 13: Y/y changes in unemployment and credit card Figure 14: Y/y changes in unemployment and credit card
charge-offs since January 2001 charge-offs since January 2007
21 December 2009 85
Barclays Capital | U.S. Securitized Products Outlook 2010
Even in early amortization, senior How credit card ABS trusts perform in a stressed collateral environment is primarily a
credit card ABS are generally function of yields, charge-offs and payment rates (and the interaction among these
well protected variables). Obviously, if a trust avoids early amortization, the noteholders will most likely be
paid in full on the expected maturity date. To this end, the support put forth by most major
credit card ABS issuers (the lone exception being Capital One) in 2009 has been beneficial
to all classes of notes; the addition of credit enhancement has allowed the trusts to
maintain their credit ratings, while the use of the discount option has improved yield and
excess spread. We believe support will continue to be forthcoming from the major issuers in
2010, should it become required to maintain ratings or avoid early amortization. However,
as the experience of Advanta’s ABCMT is bearing out, trusts that fail to be supported and
fall into early amortization generally suffer losses on lower-rated classes. Senior noteholders
tend to fare better, as in the NextCard and First Consumers cases, and we believe ABCMT
senior noteholders will likely be repaid in full by year-end.
The spread rally is likely to Like the auto sector, credit card ABS spreads have been tightening since late 2008. The
continue, provided there is uncertainty introduced by implementation of FAS 166/167 and the consequent FDIC safe
favorable FDIC safe harbor harbor risk in late 2009 did little to slow this trend; spreads widened briefly at the end of
resolution November, but recovered as investors put sidelined cash to work ahead of the new year. In
2010, we believe mezzanine and subordinate credit card ABS represent good relative value,
as they appear to us to have the most room for further tightening. The limited new issue
volume we expect, combined with stabilizing collateral performance, ought to keep the
tightening trend intact. Of course, this assumes a favorable resolution to the FDIC safe
harbor and legislative initiatives to attach new requirements to securitization.
Figure 15: Credit card ABS spreads tightened significantly over the year
3500
3yr AAA 3yr A 3yr BBB
3000
2500
2000
1500
1000
500
0
Jan-07 May-07 Sep-07 Jan-08 May-08 Sep-08 Jan-09 May-09 Sep-09
Source: Barclays Capital
21 December 2009 86
Barclays Capital | U.S. Securitized Products Outlook 2010
Issuance volume likely to remain With the assistance of the TALF program, student loan ABS issuance in 2009 totaled
muted $18.0bn, down 36% from $28.1bn in 2008 and $62.3bn in 2007. The decline is largely due
to Department of Education (ED) programs (its loan purchase program and the Straight A
Funding conduit) put in place in 2008-09 to finance student lending for the 2007-08, 2008-
09, and 2009-10 academic years in lieu of the seized ABS market. These programs sucked
up substantially all new student loan originations and, thus, potential securitization
collateral. Nevertheless, 14 transactions from three issuers (SLM Corp, Nelnet, and Student
Loan Corp) were brought to market. Approximately $8.9bn was backed by private student
loans, $7.4bn of which was TALF eligible; however, none of the FFELP issuance was. As with
other consumer ABS, issuance would likely have been minimal without the benefit of TALF.
We expect limited student loan ABS issuance in 2010 yet again, due to the continuation of
alternative funding programs and the likely elimination of FFELP (the House passed a bill
eliminating the program; action is yet to come in the Senate). Nevertheless, we expect some
clean-up transactions of seasoned FFELP loans held on balance sheet and the possible
securitization of loans financed through the ED’s purchase program and/or Straight A
Funding conduit, if market conditions are favorable. As such, our estimate for 2010 student
loan ABS volume is $15-20bn.
Figure 16: Student loan ABS volumes declined as other funding sources ramped up
50
40
30
20
10
0
2001 2002 2003 2004 2005 2006 2007 2008 2009
Source: IFR Markets
Collateral performance The national cohort default rate for FFELP loans, published approximately two years after
deteriorated with the economy, the fiscal year in which the respective students enter repayment, increased in 2007 to 6.7%
but should stabilize in 2010 from 5.2% in 2006 and 4.6% in 2005. This is the highest level since 1998 and reflects the
weakening economy and credit crisis of the past two years. Historically, student loan
defaults have been relatively low, largely because student loan borrowers have more flexible
repayment options, which include deferment, forbearance, and consolidation. The strong
performance of student loan collateral during periods when default rates picked up on other
consumer asset classes can be attributed to improved debt management counseling by
loan holders, who encouraged borrowers to use deferment and forbearance. In addition,
private and FFELP student loans are non-dischargeable in the event of bankruptcy, adding
another layer of protection for the loan holder and, consequently, ABS investors.
Spreads on student loan ABS declined in 2009, trading in relationship to other consumer
ABS sectors. We expect FFELP ABS spreads to continue to tighten in 2010. Limited supply
and the potential elimination of FFELP could lead to a spread rally on scarcity value, but we
believe other consumer asset classes have more potential for tightening. The private
student loan sector could present interesting opportunities for investors willing to do the
21 December 2009 87
Barclays Capital | U.S. Securitized Products Outlook 2010
credit work. As with FFELP ABS, private student loan ABS woefully lack for data. Investors
must sift through mountains of remittance reports to tease out default, prepayment, and
recovery rates. However, those willing to dig into the data and deal documentation stand to
reap rewards in the niche sector.
Figure 17: Student loan ABS spreads trended lower, following other consumer ABS
600
1yr 3yr 5yr 7yr
500
400
300
200
100
-100
Jan-07 May-07 Sep-07 Jan-08 May-08 Sep-08 Jan-09 May-09 Sep-09
Source: Barclays Capital
21 December 2009 88
Barclays Capital | U.S. Securitized Products Outlook 2010
SPECIAL TOPIC
While our base case of a 35% peak-to-trough (P2T) in national CS HPA is essentially
unchanged, extended foreclosure paralysis has substantially increased the likelihood of a
delayed bottom. We present a delay scenario in which home prices bottom at 34% P2T, but
not until Q2 11. Offsetting this is a strong recovery scenario in which home prices have
already bottomed. We discuss the causative agents and assumptions that lead to each
scenario, as well as HPA prospects in the bust states.
Our regional HPA model incorporates a variety of new factors. With an increasingly deep set
of home price data, we can refine our estimations of the above effects, as well as of
unemployment. The basic variables that enter our HPA regression are abnormal seasonals,
REO levels and changes, unemployment changes, and rates/tax incentive. After discussing
the key ingredients of the model, we briefly outline our calibration methodology, apply the
model at the state level to various scenarios, and, finally, aggregate to the national level.
Abnormal seasonals
Distresed share of sales We hypothesized an abnormally strong seasonal effect due to fluctuating distressed shares
fluctuates seasonally, leading to in seasonal HPA biases (Securitized Products Weekly, July 31, 2009). Our basic observation
distortions in HPA was that voluntary sales vary by nearly a factor of two between winter and spring, whereas
distressed sales vary only 30-40%. Thus, the distressed share of sales oscillates strongly
between summer and winter, affecting home price indices due to compositional and
geographic shifts. We estimated that annualized HPA receives a roughly 8-13% additive or
subtractive effect in summer or winter due to such effects.
Abnormal seasonals alone We recently examined the First American Core Logic’s Loan Performance HPI series, which
suggest a reduction in HPA includes a sub-series in which distressed sales are excluded. These new data confirm our
of 20 points, compared hypothesis, as well as the magnitude of the effect. Figure 1 shows the difference between
with summer peaks annualized HPA using the full data set and the distressed-excluded set for each of the 50
states. The line shows the weighted average for the US. Because abnormal seasonals arise
from distressed shares, the difference should show very little seasonal variation until the
past couple of years, when REO inventories skyrocketed. The data do follow this pattern, as
well as the predicted dip each winter and a boost each summer. On average, the distressed-
excluded index shows stronger HPA, by about 2%, as expected due to the non-seasonal
component of distressed share. Superimposed on this is an approximately 10% seasonal
effect. Thus, the “all” index averages 2% weaker HPA than the distressed-excluded index
overall, but was 8% stronger in summer 2009. Assuming the distressed-excluded HPA was
unchanged, the “all” HPA measurement would be 12% lower than the distressed-excluded
21 December 2009 89
Barclays Capital | U.S. Securitized Products Outlook 2010
HPA in winter 2010, if this seasonal pattern holds. That is 20% weaker HPA than has
recently been displayed.
To incorporate this effect into regional fits, we include an independent variable that is
simply the product of a standard seasonality vector (based on pre-crisis variation) with the
level of distressed inventories. The rationale is that abnormal seasonals apply only when
REO inventories become large.
Distressed Inventories
The crux of our HPA/REO model is the depressive effect foreclosure sales have on nearby
home prices. Early in the housing crisis, we could draw only on sparse academic research13
to postulate a negative feedback look (Subprime challenges: Housing and foreclosures,
February 28, 2008). As distressed inventories mounted, it became possible to calibrate the
temporal dependence of HPA on REO using geographic panel data (California REO-HPA
dynamics, October 8, 2008). With another year of data available, we can now dispense with
these proxies and calibrate directly using time-series data.
Our HPA/REO regressions We regress at the state level, using the level of and one-month change in distressed
suggest that REO change leads inventories (expressed as a percentage of housing stock). A supply-demand equilibrium
HPA by about 3 months… picture suggests that HPA should be proportional to changes in REO. However, a servicer in
a non-equilibrium environment will certainly consider the size of his or her portfolio when
deciding how to price. Empirically, REO change is the larger short-term driver, leading HPA
by about 3 months on average (Figure 2 shows our benchmark example, California, with
REO change scaled by -250). Because roll rates and existing delinquency pipelines allow us
to forecast REO years in the future (although servicer behavior introduces considerable
uncertainly over a period of a few months), shadow inventory has a dominating influence
Source: First American Core Logic’s Loan Performance HPI, Barclays Capital
13
The external costs of foreclosure: The impact of single-family mortgage foreclosures on property values,
Immergluck and Smith, Housing Policy Debate, 2006.
21 December 2009 90
Barclays Capital | U.S. Securitized Products Outlook 2010
on our HPA projections. For example, servicer delays in foreclosure can push REO
inventories off by a few months, but because seasonals are so strong in spring, this can
sometimes move a forecasted bottom to the next period of seasonal weakness, the
following winter.
Figure 2: CA HPA and distressed inventories Figure 3: CA HPA, unemployment, and rate attractiveness
21 December 2009 91
Barclays Capital | U.S. Securitized Products Outlook 2010
change is the expansion to certain existing homeowners. Existing home sales imply that
about 68% of homeowners have resided at least five years, making them potentially eligible
for the credit. However, most of these purchases will also sell their existing home, reducing
the net effect on home prices. In a balanced environment in which buyers and sellers are
equally matched, the tax savings might be equally shared, reducing the effect of the
purchase incentive by a factor of two. However, with months’ supply hovering around 7
(compared with a historical norm closer to 5), we estimate a reduction of 3.4. This leads to
an incremental rate-equivalent change of 18bp. Thus, we expect the effect of the tax
incentive expansion to be about 50% of the effect of the original program.
Incorporating tax incentives into an effective mortgage rate leads to an adjusted rate
attractiveness measure (Figure 3). The 100bp rally at the end of 2008 combined with the
initiation of the tax incentive to cause adjusted rate attractiveness to spike just prior to the
improvement in HPA, and subsequent fits attribute a roughly 5% improvement in mid-2009
due to this effect.
Unemployment
Unemployment was a lagging Figure 3 shows unemployment rates in California, but the general observations are echoed
indicator for housing in 2008-09, across other bust states. Unemployment rose only after the housing component of the
though that could change economy collapsed (housing construction, residential investment,, making it a lagging
indicator. On the other hand, it spiked sharply and in coincidence with HPD in early 2009,
before stabilizing in mid-2009.
So far in this housing downturn, unemployment has been strongly correlated with HPA.
This may change, so it is important to gauge its effect. In prior busts, unemployment was a
primary driver of HPA, and our views on the key historical drivers affect the calibration to
unemployment. Thus, the degree to which we attribute HPA to unemployment differs
depending on the level of REO inventory.
Methodology
We use First American Core Logic’s Loan Performance HPI non-seasonally adjusted home
prices to calibrate our model at the state level. The basic functional form of the regression is
where REO is expressed as a percentage of housing stock (eg, California is at 1.1%), dRdt is
the monthly change in REO, lagged 3 months, dUdt is the 3-month change in
unemployment rate (eg, California rose from 11.9 in July to 12.5 in October, so dUdt was 0.6
in October), RA is adjusted rate attractiveness expressed in bp (116 in October), and SV is a
state-dependent 12-month seasonality vector.
In bust states, we believe distressed inventories are the primary drivers of HPA. To express this
view, we first regress HPA against REO-related variables, then fit the residual to
unemployment and rate attractiveness. This is equivalent to choosing the components of dUdt
and RA that are orthogonal to REO and dRdt. For each state, we also do the reverse, fitting to
non-REO variables first and using REO only secondarily. This generally results in larger betas
on unemployment and is appropriate in stable states without heavy distressed inventories.
Finally, we interpolate between the “bust” and “stable” regressions based on REO. The
interpolation function14 is constructed so the fit is 80% “bust” in California, but 80% “stable”
14
Logit(x), where x = (REO – 0.75)/.25.
21 December 2009 92
Barclays Capital | U.S. Securitized Products Outlook 2010
in a states such as Texas (REO = .4%). A natural consequence of this interpolation is that even
bust states are increasingly treated like stable states as REO inventories decline.
Our updated HPA model trims An additional methodological twist is that we censor outlier betas. For example, we expect HPA
parameters that are far to depend on rate attractiveness in roughly the same way across each state, so the beta to that
from the national mean variable (ra) should not vary by multiple factors. To balance state-specific demographics and
dynamics with uniformity across the country, we smoothly trim parameters that are far from
the national mean. Algorithmically, this is done with a logit function that penalizes parameters
that deviate by more than a standard deviation. This trimming procedure helps ensure sensible
signs on fit parameters (eg, HPA should improve if rates fall).
Our HPA model does reasonably Over our calibration period (mid-2005 on, to avoid most of the housing boom), historical
well in back-fits, including fits for larger and bust states typically display R-squareds of 65-80%. This simply means the
predicting the 2009 upturn in fits are good, not that they have predictive power. However, calibrations that end prior to
home prices 2009 do a surprisingly good job predicting the upturn in 2009. Fit projections almost always
get the timing and direction right, although they can sometimes miss the magnitude of the
improvement by a factor of two. We did not expect even this degree of success out-of-
sample, given that the dynamics in 2009 included many effects that were absent in prior
years. We hope, of course, that the model will fare better now that it is calibrated to both
deteriorating and recovering HPA environments.
California
Our projections for California (Figure 5) suggest a 15% further decline in prices in the base
case, levelling off in early 2011. The regression actually indicates a very slight further decline
in 2012, but given uncertainty about the environment so far out, we consider early 2011 the
effective bottom. In the stress case, California home prices fall an additional 23%,
highlighting the risk of enormous shadow inventory if it cascades into supply swiftly.
Interestingly, the optimistic scenario shows prices appreciating in 2010, only to dip a bit in
2011. This reflects the prolonged build-up of shadow inventory, which needs to dissipate
eventually. In the optimistic scenario, inventory continues to weigh on HPA for several years,
pressuring prices as historically low rates and the effects of tax incentives wear off. While we
can certainly envision strong rebounds associated with economic vitality, the cautionary
lesson is that delaying foreclosures is not the same as eliminating them, and heavy pending
supply could make post-bust housing markets vulnerable to even minor downturns.
21 December 2009 93
Barclays Capital | U.S. Securitized Products Outlook 2010
100
1.4
95
1.2
90
1.0 85
stress 80 stress
0.8 base base
75
optimistic optimistic
0.6 70
Oct-09 Oct-10 Oct-11 Oct-09 Oct-10 Oct-11
Source: Barclays Capital Source: Barclays Capital
Our state-level HPA projections When we aggregate state-level projections to the US, the base case corresponds to 8%
aggregate to an 8% remaining remaining depreciation (Figure 7), with recovery in Q2 10. Thus, our US forecast is nearly
drop at the national level unchanged at 35% peak-to-trough (for CS). A scenario (not shown) in which foreclosure
paralysis remains for an additional quarter before F2R grows to 5% also projects 8%
remaining, but with a bottom in Q1 11. Thus, the timing of a housing recovery is sensitive to
the behavior of servicers. A short foreclosure hiatus could push the distressed inventory
problem into next winter. We would distinguish this scenario from our optimistic one by
tracking REO liquidations, as the optimistic scenario depends on surging demand. In that
case, home prices already bottomed in Q2 09. The stress scenario shows a 14% decline,
corresponding to 38% peak-to-trough, representing the continued vulnerability to a
shadow inventory shock.
Our model does not incorporate All our projections assume mortgage rates remain constant. In fact, we expect rates to rise
a rise in mortgage rates, but we in 2010, with mortgage rates selling off 100bp or more. However, this expectation is
expect any such rise to be offset founded in a recovery environment in which consumer confidence should lead to greater
by higher demand in a housing demand, not weaker. Thus, while our regressions suggest that higher mortgage
recovering economy rates could decrease HPA by roughly 5%, such an outcome should be compensated by
higher demand.
In our view, home prices will depreciate in Q4 09 and Q1 10. Assuming REO inventories
continue to rise in the short term, we expect an 8% decline to end in Q2 10. If servicers
delay foreclosures, the bottom may be pushed to early 2011, because strong seasonals
21 December 2009 94
Barclays Capital | U.S. Securitized Products Outlook 2010
HPI
St Stress Base Optimistic 105
AZ 22 17.3 6.2
100
CA 22.7 15.3 7
FL 22.4 12.2 2.2 95
IL 13.8 10.8 2.7
MI 24.9 19.4 16 90
NV 34.5 30.4 27.3
85 stress
NY 13.4 6 0.2
OH 7.2 4.9 0.5 base
80
PA 1.7 1.7 -0.4 optimistic
75
Nov-09 Nov-10 Nov-11 Nov-12
Source: Barclays Capital Source: Barclays Capital
should support HPA in Q2 and Q3 10. The remaining downturn will be more severe in the
bust states, we believe, because distressed inventory is large and pending, not removed.
The consequences of delayed foreclosures must still be accounted for, even in an
environment with recovering demand.
21 December 2009 95
Barclays Capital | U.S. Securitized Products Outlook 2010
SPECIAL TOPIC
Starting in H2 10, there could be New mortgage origination is essentially limited to the GSE and FHA channels, both of which
a meaningful extension of have tightened credit dramatically over the past two years. This has contributed to flat
conventional credit to mid tier agency and non-agency refinance curves and the near-elimination of credit curing in Alt-A
FICO borrowers and subprime deals. It also presents a drag on the housing sector, where distressed supply
continues to weigh on the market. In this article, we look at the underlying causes of the
current tight credit conditions and argue that most such issues will become less important
over the course of 2010.
Agency credit standards have tightened sharply over the past few years
The agency credit box has To formulate a view of the availability of mortgage credit, it is important to review what has
tightened substantially over the taken place over the past few years, as well as to paint a better picture of the current credit
last couple years landscape. With the growth in the non-agency lending market in 2006-07, the GSEs relaxed
their underwriting guidelines significantly. Yet as housing started to deteriorate and
delinquencies began to ramp up in the GSEs’ guarantee books, the GSEs shifted to a much
more conservative posture. For example, average FICO scores for new Freddie Mac
originations increased from about 720 to 760 from mid-2007 to today (Figure 1). At the
Figure 1: FICO on new originations have improved Figure 2: Credit to lower FICO borrowers has been cut
Source: HUD, Freddie Mac, Barclays Capital Source: Freddie Mac, Barclays Capital
21 December 2009 96
Barclays Capital | U.S. Securitized Products Outlook 2010
same time, the percentage of loans with original LTV >80 also dropped from a high of more
than 35% down to about 17% currently.
Because of this shift to tighter underwriting standards, many seasoned borrowers who
originally qualified for credit under more lax standards have now been effectively locked out
of new credit. In Figure 2, we look at the distribution of FICO scores for the Freddie Mac
2007 and 2009 vintages. Not only is the average credit score for the 2009 vintage
significantly higher, but the distribution is skewed much more toward higher scores. For
example, more than 50% of 2007 borrowers had a FICO of <720. In contrast, only 10% of
the 2009 vintage did.
Significant underwriting Still, the FHA program has generally tightened credit to a much lesser extent than FN/FR. For
changes are in the works for example, while Fannie Mae and Freddie Mac were implementing loan-level pricing
the FHA program adjustments based on the exact risk characteristics of individual loans (more on this topic
later), the FHA program placed a moratorium on risk-based pricing. Only more recently has it
begun to step up its efforts to tighten underwriting standards. For example, in September, the
FHA announced sweeping changes to its streamlined refinancing program.15 More recently, it
announced plans to tighten credit in 2010 more broadly through a variety of mechanisms. As
these changes have the potential to affect credit and subsequently borrower and prepayment
behavior significantly, we summarize some of the major proposals.16
Increase minimum FICO score: No exact level has as yet been suggested other than
stating that the “relationship between FICO scores and down-payments” is being
investigated.
Increase down-payment requirement: Most FHA loans require 3.5%. By increasing the
down-payment requirement, borrowers would effectively have more equity and, thus,
lower LTVs.
Increase the mortgage insurance premium (MIP): The FHA is exploring increasing the
MIP required by borrowers.
Increased lender accountability: The FHA would like to increase lender responsibility
for loans that are not underwritten the FHA standards. The FHA has proposed posting
a Lender Scorecard to increase “transparency and accountability.”
15
For more information, please see “Prepayment Commentary” Securitized Products Weekly, September 25, 2009.
16
For more information, please see “Trends and Issues” Securitized Products Weekly December 4, 2009.
21 December 2009 97
Barclays Capital | U.S. Securitized Products Outlook 2010
Loan level pricing adjustments While a comprehensive analysis of all GSE underwriting changes is beyond the scope of this
seem to be only a minor factor article, there are several key points worth highlighting in examining the their role in tighter
reducing credit availability for lending standards. One of the most visible factors that have been associated with tighter
mid-tier borrowers GSE underwriting has been the introduction of loan-level pricing adjustments, or LLPAs.
These fees can be as high as 3 points upfront (Figure 3), which could add as much as 75bp
to a borrower’s mortgage rate. While this certainly is going to dampen the economic
incentive of some borrowers to refinance, it does not explain the reduction in refinancability
of borrowers in the < 80 LTV, 700-740 FICO bucket. For these borrowers, LLPAs range from
25bp to a point and, thus, should only add 25bp maximum to their mortgage rate. Given
historically low mortgage rates, an additional 25bp does not seem to be a material
difference. While LLPAs are a contributor, they do not seem to be the main factor leading to
a more restrictive underwriting box.
For manually underwritten mortgages, when the borrower’s monthly debt payment-
to-income ratio exceeds 36%, the seller must document the justification for the
higher qualifying ratio.
Increasing documentation requirements has also served to reduce the credit available for
mortgage borrowers. For example, as we show in Figure 4, documentation changes
Figure 3: LLPAs have increased the cost of refinancing Figure 4: Low documentation options have gone away
>60% & >70% & >75% & >80% &
35
Credit Score < 60% <=70% <=75% <=80% <=85% > 85%
≥ 740 0.00% 0.25% 0.25% 0.25% 0.25% 0.25%
30
≥ 720 & < 740 0.00% 0.25% 0.25% 0.50% 0.25% 0.25%
25
≥ 700 & < 720 0.00% 0.75% 0.75% 1.00% 0.75% 0.75%
20
≥ 680 & < 700 0.25% 0.75% 1.25% 1.75% 1.25% 1.00%
15
≥ 660 & < 680 0.25% 1.25% 2.25% 2.75% 2.50% 2.00%
10
≥ 640 & < 660 0.75% 1.50% 2.75% 3.00% 3.00% 2.50%
5
<640 0.75% 1.75% 3.00% 3.00% 3.00% 3.00%
0
Jan-06 Oct-06 Jul-07 Apr-08 Jan-09 Oct-09
Stated asset Stated income No VOE
Source: Freddie Mac, Barclays Capital Source: Freddie Mac, Barclays Capital
17
http://www.freddiemac.com/sell/guide/bulletins/pdf/bll112408.pdf
21 December 2009 98
Barclays Capital | U.S. Securitized Products Outlook 2010
announced in late 2008 essentially ended GSE limited documentation originations, which
accounted for as much as 30% of all originations during 2006-07. While this has limited the
credit available for low doc borrowers, the biggest effect likely comes from the incentive
they create for lenders to tighten credit. Over the past couple of years, increased
documentation requirements for appraisals, income, and other areas have raised the burden
on originators. Lenders must insure compliance with all documentation guidelines, or else
risk having loans put back to them in the event of default. We will further address originator
repurchase risk below.
Debt-to-income limits seem A significant shift, in our view, has arisen from the changes to GSE maximum debt-to-
unable to explain the upward income requirements to 45%. As we show in Figure 5, this led to a large reduction in debt-
shift in FICO to-income ratios for new GSE originations, from 38% to 32% in aggregate. At first glance, it
would seem that this is responsible for the sharp increase in FICO scores for new
originations. However, the numbers below suggest that the effect is much smaller than it
first appears. In Q3 08, the average FICO for new originations with DTI less than 45% was
742 compared with 729 for DTI greater than 45%. At the same time, borrowers with DTI
greater than 45% represented only roughly one-third of Q3 08 originations. Thus, even if
new originations cease to have borrowers greater than 45 DTI, this could explain only about
4-5 points of the improvement in FICO scores. Even after controlling for DTI as we show in
Figure 6, there was a uniform 20 point increase in FICO between Q3 08 and Q2 09. Thus,
while underwriting changes to DTI can explain some of the changes to FICO, they do not
seem to be the main driver in credit tightening.
We estimate that GSEs are Nor do the GSEs seem to be. We have shown that although they have sharply tightened DTI
responsible for a third of credit requirements starting in 2009, this does not explain most of the credit tightening during the
tightening over the last year past year. Overall, we estimate that they could be responsible for roughly one-third of credit
tightening in the agency sector. It is difficult to quantify how they will influence future
underwriting standards. They have to balance the political pressure to make mortgage
credit available, while at the same time reducing their credit risk on their ongoing book of
business. That being said, it seems unlikely that they will loosen credit over the next year,
unless they are given a clear mandate from the administration to do so. However, with the
utter failure of HARP and the disappointing results thus far of the HAMP program, we would
not discount this outcome.
Figure 5: 30y fixed-rate debt-to-income ratio (%) Figure 6: DTI alone does not explain the upward drift in FICO
40 FICO
39 790
780
38
770
37
760
36 750
35 740
34 730
33 720
32 710
31 700
690
30
1 5 9 13 17 21 25 29 33 37 41 45 49 53 57 61 65
Jan-06 Oct-06 Jul-07 Apr-08 Jan-09 Oct-09
Orig DTI
DTI 2008q3 2009q2
Source: Freddie Mac, Barclays Capital Source: Freddie Mac, Barclays Capital
21 December 2009 99
Barclays Capital | U.S. Securitized Products Outlook 2010
Repurchases
Repurchase risk seems to be Perhaps the greatest worry facing underwriters is the possibility that they will be forced to
the biggest factor driving credit repurchase poorly underwritten delinquent loans. During the past two years, bank loan
tightening repurchases for violations of reps and warrantees have increased markedly (Figure 7).18 In
Q3 09 alone, banks have been obligated to repurchase nearly $8bn in non-performing loans,
bringing their YTD total to nearly $14bn. Assuming banks mark repurchased loans at $0.40
on the dollar, loan repurchases may cost banks well over $11bn in 2009. It is important to
note that these repurchases represent primarily seasoned originations (especially the 2006-
07 vintages), not new issuance.
As systems and staff are There are three reasons to expect far smaller repurchase rates on new origination. First, the
retooled, originators should be economic environment should be more robust than the 2007-09 housing and credit crisis.
able to relax harsh FICO and Second, delinquency rates on less leveraged loans are much lower than those on high-LTV
documentation requirements and/or affordability products. Even if new origination encompasses mid-FICO borrowers,
and even if the economy double-dips, default rates should be far smaller than the 2009
experience, as long as substantial down-payments and sensible mortgage products are the
norm. FNMA guidelines state that loans that go delinquent in the early years may be subject
to an underwriting review, and lenders may be required to repurchase such loans if they
have “significant underwriting deficiencies.” Thus, repurchases can be minimized by
ensuring lower delinquencies (low leverage), but also by documenting loans zealously.
Anecdotal evidence strongly suggests that many repurchases are the result of poorly
documented loan files. For example, a loan may be LP or DU accepted pending verification of
income, but if the file is missing the VOI, a repurchase could result. Thus, the third reason for
smaller repurchases rates on new origination is far better documentation and quality control.
In the face of massive repurchases, underwriters should naturally move to very high FICO
borrowers and demand far greater documentation. Once systems and staff are retooled/
retrained to ensure consistent documentation, however, originators should be able to relax
Figure 7: Bank loan repurchases have risen sharply ($bn) Figure 8: Large banks have been hit by repurchases ($bn)
9 6
8
5
7
6 4
5
3
4
3 2
2 1
1
-
-
2008
2009
2008
2009
2008
2009
2008
2009
2008
2009
Q1 Q2 Q3 Q4 Q1 Q2 Q3
Source: Federal Reserve, Barclays Capital Source: Federal Reserve, Barclays Capital
unduly harsh FICO and documentation requirements. Given multiple platforms and
18
As a result of its 1-4 family residential mortgage banking activities, a bank holding company may be obligated to
repurchase mortgage loans that it has sold or otherwise indemnify the loan purchaser against loss because of
borrower defaults, loan defects, other breaches of representations and warranties, or for other reasons. Repurchased
1-4 family residential mortgage loans include those that the bank holding company had sold but subsequently
repurchased under provisions of the sales agreement because of a delinquency, noncompliance with the sellers’
representations and warranties, fraud or misrepresentation, or any other repurchase requirement.
thousands of employees, such a refitting could take several months. We believe that quality
assurance can be achieved at many banks within about six months, at which point
repurchase risk should be a much weaker constraint on new origination.
Risk aversion
Having skirted the brink of catastrophe, risk managers are justifiably concerned about
taking on new risk in origination. Regulators look at banks laden with non-performing
assets and similarly caution against aggressive underwriting. Equity investors also look
carefully at loan loss provisions and expected credit performance. These concerns are
particularly justified when housing and a robust economic recovery remain on uncertain
ground. But risk tolerance ultimately needs to be grounded in sound economic
assessments. In our view (see “Housing risks and prospects”), home prices will show
renewed strength in mid-2010, and fears of a double-dip will continue to recede as the year
progresses. If additional signs of stability indeed emerge in 2010, heightened risk aversion
should gradually give way to a careful estimation of economic risks and benefits.
Risk aversion in the origination It will likely be years before bank non-performing assets run off, but loss expectations – and
process should wane as banks’ ability to earn their way out of them – should become increasingly clear. Origination
housing prices stabilise volumes are an important means to offset legacy losses, as long as the economics are
justified. And if mid-FICO originations make economic sense (which we discuss in the next
section), lenders with relatively few non-performing assets can use that demographic to
increase market share, which is not high on bank’s concerns at the moment, but should
gain prominence in a recovery environment.
Similarly, bank regulator queasiness on mid-FICO borrowers may well be more than offset
by the administration’s avowed aim to restore credit availability. Recent talk has centered on
small businesses, but housing is clearly a top-line focus item. As distressed supply builds in
the housing market, credit availability will be critical to maintaining adequate demand.
An originator’s profit is largely contained in the servicing rights, since origination fees
compensate for processing, and the underlying loan can typically be sold near par. Mid-
FICO loans have higher credit costs because of the risk of repurchase, as well as potential
losses from servicing a delinquent loan. On the other hand, their more benign prepayment
profile makes the MSR more valuable. A simple estimation shows that the benefit of less
negative convexity substantially outweighs the credit cost.
The costs of credit should be The cost is slightly higher if we assume an additional drag from servicing delinquent loans.
substantially lower for new Assuming $1000/loan servicing cost on delinquent loans19 and an average balance of
originations $200,000, this adds another 7% x 0.5% = 3.5bp, bringing the total estimated credit cost to
19.5bp. On a 25bp servicing strip, this is equivalent to a 0.78 reduction in the MSR multiple.
However, this is a worse-case scenario, assuming a repeat of 2009 credit conditions, lax
underwriting standards (80/20s, 10/20 IOs, etc.), and poor servicer documentation of those
standards. As Figure 9 shows, delinquency rates in 2008 were far lower, leading to an
estimated credit reduction in the MSR multiple of only 0.2, less than a third of the 2009 cost.
Delinquencies in 2007 were even lower, by an additional factor of five.
The convexity benefits for mid We conclude that mid-FICO origination has a worst-case increase in marginal credit cost of
FICO loans should overwhelm about 0.78 MSR. Given the likelihood of a stronger economy (eg, conservatively weighing
the credit costs the probability of a 3-year forward environment mirroring the 2008 or 2009 experiences at
60/40%), a more realistic expectation is 0.4. Factoring in lower LTVs and more sensible
mortgage products would lower this even more. However, the big change is more rigorous
Figure 9: Prime % 60+ delinquent loans in Q3 07 to Q3 08 Figure 10: Prime % 60+ delinquent loans in Q3 08 to Q3 09
15 15
FICO 650 - 700 FICO 650 - 700
FICO 700 - 750
FICO 700 - 750
12 12 FICO 750 - 800
FICO 750 - 800
9 9
6 6
3 3
0 0
0 6 12 18 24 30 36 42 0 6 12 18 24 30 36 42
19
We assume $500/year for two years; see Congressional oversight report, October 11, 2009, Pub. L. No. 110-343, pg 69.
documentation. After all, a homeowner who defaults because of job loss does not
precipitate a repurchase, as long as the originator can show proper paperwork and
underwriting standards. In our view, rigorous quality control could lower repurchase rates
by half, bringing the incremental credit cost of mid-FICO origination to 0.2 or less. As we
shall see in the next section, this is small potatoes compared with the boost in multiple due
to superior convexity.
One way to quantify the effect on MSR valuations is to examine the effect of relative FICO
on refinancing multiples. Or put more simply, how the refinancing response of specific FICO
borrower changes as a result of a shift in the FICO of new originations.20 Figure 12 shows
refinancing multiples by FICO for the 2009 and 2008 agency originations. For 2009, a
relative FICO of -50 signifies a borrower whose credit score was 712, or 50 points less than
the average for 09 originations. The data show that, on average, these borrowers refinanced
at only 60% the efficiency of 762 FICO borrowers during 2009.21
Figure 11: Lower FICO loans have a convexity advantage Figure 12: Refinancing multiples by FICO
40 1.2
1
30
0.8
20 0.6
10 0.4
0.2
0
0 25 50 75 100 125 150 175 200 0
-125 -100 -75 -50 -25 0 25 50 75
Refinancing Incentive
Relative FICO
< 700 700-730 730-760
760-790 790-820 2008 2009
Source: Freddie Mac, Barclays Capital Source: Freddie Mac, Barclays Capital
20
We measure relative FICO multiples as the prepayment ratio, after adjusting for turnover, between 12-30 WALA, 50-
100 bps ITM loans during the course of a year. Relative FICO is measured using the average FICO of new originations
as a reference.
21
The dependence of refinancing multiples on relative FICO is not constant over time and depends on other factors
such as HPA, credit availability, and economic conditions. Currently, the dependence is near historical highs. For
example, during the 2002 refi wave a lloan with a relative FICO of -50 had a 80% multiple, as opposed to 60% in 2009.
We estimate a convexity benefit Now that we have estimated the relative prepayment advantage of mid-FICO loans, we can
of 0.6-0.7 higher MSR multiple show the benefits to MSR valuations. In Figure 13, we examine valuations of mortgage
for low FICO loans relative to a servicing rights (we are using a 5% trust IO as a proxy for MSR valuations and ignore
credit cost of 0.2 MSR multiple or ancillary income), using an even OAS approach under various model refinancing multipliers.
less For servicing rights off an average 2009 loan, 775 median FICO, our model suggests a MSR
multiple of 4.17. However, for a 725 FICO loan, assuming a model refinancing multiplier of
60%, the MSR multiple is 4.82, which is roughly 0.6-0.7 higher. Relative to the incremental
credit costs we estimated of a 0.2 multiple or less, the convexity benefits of originating
moderate FICO loans trump the increase in credit costs.
As credit loosens, mid-FICO S-curves should gradually steepen, and the prepayment
multiple could trend higher. Even at an 80% multiple, however, convexity costs outweigh
credit. In addition, an easier credit environment with faster prepayments should reduce
credit costs further. Compared with present high-FICO originations, mid-FICO originations
seem even more sensible economically.
Prospects
Starting in H2 2010, we expect Agency credit has tightened sharply over the past two years and remains restrictive even as
underwriting criteria to economic conditions improve. Both the GSEs and banks need time to access the magnitude
gradually normalise of legacy loan losses and repurchases. FNMA and FHLMC have shrunk their credit box, but
that is only part of the reason for tight credit. Faced with a wave of repurchases, banks are
reacting by demanding pristine credit, low LTVs, and stringent documentation. As more
systematic documentation and quality assurance programs are implemented, they should
find themselves positioned to open the credit box incrementally. Logically, mid-FICO
borrowers (700-740) should be the first recipients. In contrast, loan-to-values above 80%
are likely to take far longer to appear, and low-doc loans may well be gone for good.
There are sound economic reasons to originate mid-FICO loans. The credit cost, by our
estimation, translates into MSR servicing multiple reductions of 0.2 or less. On the other hand,
the flatter prepayment profile of these borrowers is worth, in our even-OAS analysis, 0.6-0.7.
Further, the convexity advantage is greater for first movers because the prepayment profile
should remain extremely flat until mid-FICO origination volumes approach historical norms.
The need to maintain origination volumes and the administration pressure to restore credit
argue for expansion to mid-FICO borrowers. This pressure will be even greater if mortgage
rates sell off, as we expect in 2010, because purchase borrowers tend to be lower credit.
Originators need time to update systems to comply with a barrage of GSE underwriting
changes, to retrain loan officers to meticulously document loan files, and to assess
expectations on losses adequately. We believe, however, that these tasks can be completed
at many banks over the next six months. Thus, we expect mid-FICO, LTV < 80% originations
to pick up over the next 6-12 months. Contingent on a continued recovery environment,
lower (680-ish) FICO borrowers are the likely subsequent beneficiaries, well before higher
LTV borrowers.
If credit expands as we expect, agency prepays should pick up on pools with high
concentrations of mid-FICO borrowers. On the other hand, 2006-07 vintages with high
proportions of high-LTV borrowers are likely to remain slow. On the non-agency side, mid-
FICO originations open the way for some credit curing in subprime pools. For low dollar-
priced securities, even limited curing can significantly boost valuations (see “Subprime
speeds: Assessing credit curing”). Because mid-FICO borrowers have been the dominant
demographic home-buying segment, credit expansion to this group is critical to the
absorption of shadow housing inventory and a long-term stabilization in home prices.
SPECIAL TOPIC
Voluntary prepayment speeds for subprime loans have fallen continuously since the peak of
the housing bubble in summer 2006. Historically, home price appreciation and easy credit
access (ie, high issuance volumes) provided subprime homeowners ample refinance
options. Now, home price depreciation has restricted options for those with LTV>100%.
Even for low LTV borrowers (LTV<80%), nonexistent subprime issuance prevents
refinancing. Worse, agency underwriting standards have risen dramatically (“Agency
underwriting: When will the squeeze ease?”), curtailing refis on even the subset of subprime
borrowers who cure to about 700 FICO. The result is that voluntary prepayments (CRRs)
have ground to a halt (Figure 1).
Investors appear to have assumed a future not very different from the current credit
environment, typically running bond cash flows assuming that recent anemic speeds
remain in place forever. Given aggregate speeds of 2.5 CRR, subprime prepay speeds are
unable to fall much lower. On the other hand, credit should naturally loosen in a recovery
environment. We view the likely catalysts for an increase in subprime CRRs to be, initially, a
renormalization of agency credit and, later, a resumption of new issue bonds characterized
by lower FICOs. We estimate the potential magnitude of improvement and determine the price
effect of credit curing on different vintages of subprime bonds. While the timing is uncertain,
we suggest that base-case speeds should be notably higher than recent prints, and
considerable upside exists for even faster speeds – making subprime an attractive asset class.
In summer 2007, when demand for subprime issuance disappeared, an important refinance
pathway was closed for existing subprime borrowers. Figure 1 shows the critical role new
issuance played in prepay speeds. Aggregate voluntary subprime speeds plunged threefold,
from 30 to 10 CRR, as non-agency issuance was extinguished. Issuance in Figure 1 reflects
our estimates for refi-purpose issuance volumes with FICO<650, expressed as a percentage
of subprime outstanding. It crudely tracks actual subprime speeds, including in H2 07, when
only FHA and GSE channels were open. Because FHA is heavily dominated by purchase
loans, our best estimate is that most subprime refinancings involve cures to FNMA or
FHLMC, rather than FHA.
Agency FICO standards should Agency standards have tightened tremendously over the past couple of years, meaning that
relax from elevated levels borrowers who previously could have refinanced into an agency product can no longer do
so. Average agency FICO scores have risen from 710 to 760 (and the median is now 780).
These FICO standards are not consistent with long run profit maximization (see “Agency
underwriting: When will the squeeze ease?”, page 97), and we expect a relaxation of credit
to 700-740 borrowers over the next 6-12 months. Such a relaxation would allow some low-
LTV subprime borrowers without prior delinquency to refinance into agency products.
50 45
40
40 35
30
30
25
20 20
15
10 10
5
0 0
Jan-06 Oct-06 Jul-07 Apr-08 Jan-09 Oct-09 Jan-06 Oct-06 Jul-07 Apr-08 Jan-09 Oct-09
Counteracting this is an expected tightening in FHA underwriting standards. Yet given that
subprime speeds now aggregate to 2.5 CRR, additional curtailment of FHA credit is likely to
have minimal effect. We estimate only about 1 CRR coming from the FHA, so even a 20%
reduction in its box would be unimportant. On the other hand, an expansion of the agency
box to H2 07 standards could accelerate speeds by a few CPR. Of course, even if credit
underwriting returns to late 2007 conditions, high LTVs and delinquency rates limit the
degree of subprime acceleration. We must therefore estimate how current LTV/delinquency
distributions affect potential pickups in speeds (see below).
Figure 3 shows the effect of LTV on prepay speed over the last four years. High LTV reduces
the ability of borrowers to refinance and can prevent sales of property, which is clearly
indicated in this chart by an inverse relationship between CRR and LTV. However, even
M2M LTV > 90 buckets prepaid faster in late 2007 than LTV < 80 borrowers are doing at
present. One reason for this is shown in Figure 2, which highlights the credit component of
prepay speeds. Borrowers who have always been current are likely to have more improved
FICO scores, and hence are more likely to refinance into a more attractive mortgage. This
time series shows not only the general contraction in credit, but how tightened lending
standards have shut out all but the most creditworthy borrowers. Credit is likely to return in
stages, with agency underwriting expanding first, followed by gradual re-establishment of
low-FICO issuance.
Low LTV subprime should return We view low-LTV, low FICO borrowers as an underserved market and remain optimistic
to at least $50 billion per year about an eventual return of low-LTV subprime-like issuance (eg, FICO of 650-700). Prior to
the boom years of 2004-07, subprime issuance for low-LTV borrowers ran at $50-70
through the late 1990s. Mortgages delinquencies primarily reflect excess leverage, through
either high LTVs, inflated incomes on low-doc loans, or strongly resetting mortgage
payments. Credit impairment matters less when sufficient equity exists as collateral. As
credit spreads narrow, we expect a return of non-agency issuance to meet investor demand
for risky assets, but with low LTVs, full documentation, and sensible mortgage products.
Ultra-clean prime deals could price in H1 10, in our view, followed by slightly lower credit
borrowers as the year progresses. In a recovery environment, we find a return of lower FICO
issuance within the next 2-3 years plausible. With subprime outstanding balances of
$800bn currently, issuance of only $50bn in refinance loans would correspond to a 6 CRR
improvement in subprime prepays.
Home price appreciation greatly Lastly, home price appreciation is a key driver of refinancing. Even a modest 10% increase
expands refi eligibility in home prices over the next several years would increase borrower eligibility for agency
loans and low-LTV non-agency loans. An expansion of credit amid an economic recovery
would be consistent with 10% HPA. Further, even a 3% US HPA number implies much
stronger HPA in many neighborhoods, and this dispersion helps prepayments (borrowers
who are driven even deeper underwater make no marginal effect on voluntary speeds, but
those who are driven below 80% LTV have new refi opportunities). In the analysis that
follows, we look at the upside potential in 2004 and 2007 vintage subprime in each of the
scenarios described.
We analyze four credit scenarios Historical prepay speeds provide a common basis of comparison across collateral attributes.
To estimate sensitivity of CRR to credit environment, we consider four scenarios: 1) prepay
speeds remain at current levels; 2) they rise to the levels of H2 07 – for each
LTV/delinquency category - over the course of 18 months, starting 6 months from now; 3)
the same as 2), followed by an additional 6 CRR due to non-agency issuance, ramping up
from month 18 to 30; and 4) the same as 3) plus home prices recover 10% over the next 30
months. Each scenario generates a unique distribution of borrowers by LTV and
delinquency status over time. To estimate the CRR effects, we evolve the collateral
distribution using current roll rates, varying only prepay speed and LTV according to each
scenario. It is important to note that each scenario is sensitive to LTV and delinquency
status. Therefore, without HPA, a high LTV borrower will not prepay rapidly in any scenario.
This procedure is probably overly conservative because roll rates are likely to fall
significantly in a recovery environment, and HPA dispersion should create more migration
into low LTV buckets than we project. Nevertheless, even with these conservative
assumptions, considerable upside potential exists.
Once we have evolved the various attributes, we fold the LTV/delinquency distribution into
the conditional prepayment speeds exhibited in each of the three credit environments. The
resulting CRR time series is used as an input to determine the value of future cash flows. One
such example is shown in Figure 5, where credit availability remains at current levels until
jumping to 2007 levels (ie, the H2 07 LTV/delinquency prepayment matrix) after one year.
Speeds slow initially as always-current borrowers fall delinquent. Note that 2004 borrowers
20% 6
15%
4
10%
5% 2
0%
0
LTV<80 LTV 80-90 LTV<80 LTV 80-90
Oct-09 Oct-10 Oct-11 Oct-12 Oct-13 Oct-14
AC CPD
Note: AC: always current; CPD: current, previously delinquent. Source: Barclays Capital
Source: LoanPerformance, Barclays Capital
receive a larger bump in prepay speeds due to their having a greater proportion of low-LTV,
always current borrowers. In the final analysis, we stagger the timing, and ramp accelerations
gradually. For six months, credit remains at current levels in each scenario. We then ramp
credit availability linearly for the next year until it reaches 2007 levels. This corresponds to an
increase in CRR for 2004 vintage subprime from 4% to 11% and for 2007 vintage subprime
from 2% to 5%, similar to Figure 5. Such an environment would only require a moderation in
agency underwriting standards, from an average of 760 to 720 FICO.
More optimistic would be a return to $50-70bn annual subprime issuance, which would
increase CRRs for both vintages by approximately 6 CRR. In our analysis, we assume this
occurs between months 18 to 30 and again take a linear interpolation of CRR over this
period. Finally, HPA has a great effect on recent vintage subprime, due to the greater
proportion of current borrowers with high LTV. For example, 10% HPA increases CRR by 2.0
for 2007 borrowers, but increases CRR by 1.2 for 2004 borrowers. Moreover, while always
current borrowers dominate prepays, we expect previously-delinquent borrowers with 2-3
year on-time payment histories to behave similarly to always current borrowers in time. We
model the HPA component as occurring between months 6 and 18. Figure 6 summarizes
these effects on voluntary prepayment speeds.
Each scenario determines a prepay vector that we apply to representative seasoned and
recent vintage bonds to illustrate and quantify the price effect. Recent vintage bonds offer
higher yields across all scenarios (Figure 7). We consider the “H2 07 conditions” scenario as
our base case. This boosts yields by 150-250bp over the “current crr” scenario. Because the
timing and magnitude of new origination is more uncertain, we view the “+new origination”
scenario as potential upside. On the other hand, a rise in HPA is very likely, in our view, so
the last scenario is perhaps even more likely than the third. This upside scenario increases
yields by 100-400bp over our base case, depending on the bond.
To estimate the price effect, we look at even-yield price changes from the yield in the
current scenario (Figure 8).
On a percentage basis, price upside in each credit scenario is comparable22, with recent
vintage bonds receiving a greater marginal benefit from HPA than 2004 bonds (up an extra
2-4% vs. up 1%). In all cases, a relaxation of credit appears to offer upside potential that is
not priced into the market. With low prepay downside, subprime offers attractive option-like
payoffs in scenarios of credit improvement.
Attractive upside A recovery environment implies lower defaults, but also faster prepayments. In our view,
exists across vintages subprime prepayment expectations are too conservative, and credit curing is likely to boost
yields significantly over “same-speed” scenarios. Home price appreciation and a re-emergence
of low-FICO issuance represent additional optionality. Given limited room for decelerations,
subprime bonds offer favourable risk-return profiles.
22
Note that the CBASS 04-CB6 AF4 is a step-up bond similar to a NAS, and actually receives principal before the AF3.
This is why the even-yield price increase is larger.
Analyst Certification(s)
We, Ajay Rajadhyaksha, Joseph Astorina, Sandeep Bordia, Glenn Boyd, Aaron Bryson, Derek Chen, Tee Chew, Sandipan Deb, Wei-Ang Lee, James Ma,
Keerthi Raghavan, Matthew Seltzer, Nicholas Strand, Robert Tayon, Jasraj Vaidya, Kumar Velayudham, and Elena Warshawsky, hereby certify (1) that the
views expressed in this research report accurately reflect our personal views about any or all of the subject securities or issuers referred to in this research
report and (2) no part of our compensation was, is or will be directly or indirectly related to the specific recommendations or views expressed in this
research report.
Important Disclosures
For current important disclosures regarding companies that are the subject of this research report, please send a written request to: Barclays Capital
Research Compliance, 745 Seventh Avenue, 17th Floor, New York, NY 10019 or refer to https://ecommerce.barcap.com/research/cgi-
bin/all/disclosuresSearch.pl or call 212-526-1072.
Barclays Capital does and seeks to do business with companies covered in its research reports. As a result, investors should be aware that Barclays
Capital may have a conflict of interest that could affect the objectivity of this report. Any reference to Barclays Capital includes its affiliates. Barclays
Capital and/or an affiliate thereof (the "firm") regularly trades, generally deals as principal and generally provides liquidity (as market maker or otherwise)
in the debt securities that are the subject of this research report (and related derivatives thereof). The firm's proprietary trading accounts may have either
a long and / or short position in such securities and / or derivative instruments, which may pose a conflict with the interests of investing customers.
Where permitted and subject to appropriate information barrier restrictions, the firm's fixed income research analysts regularly interact with its trading
desk personnel to determine current prices of fixed income securities. The firm's fixed income research analyst(s) receive compensation based on various
factors including, but not limited to, the quality of their work, the overall performance of the firm (including the profitability of the investment banking
department), the profitability and revenues of the Fixed Income Division and the outstanding principal amount and trading value of, the profitability of,
and the potential interest of the firms investing clients in research with respect to, the asset class covered by the analyst. To the extent that any historical
pricing information was obtained from Barclays Capital trading desks, the firm makes no representation that it is accurate or complete. All levels, prices
and spreads are historical and do not represent current market levels, prices or spreads, some or all of which may have changed since the publication of
this document. Barclays Capital produces a variety of research products including, but not limited to, fundamental analysis, equity-linked analysis,
quantitative analysis, and trade ideas. Recommendations contained in one type of research product may differ from recommendations contained in
other types of research products, whether as a result of differing time horizons, methodologies, or otherwise.
This publication has been prepared by Barclays Capital, the investment banking division of Barclays Bank PLC, and/or one or more of its affiliates as
provided below. This publication is provided to you for information purposes only. Prices shown in this publication are indicative and Barclays Capital is
not offering to buy or sell or soliciting offers to buy or sell any financial instrument. Other than disclosures relating to Barclays Capital, the information
contained in this publication has been obtained from sources that Barclays Capital believes to be reliable, but Barclays Capital does not represent or
warrant that it is accurate or complete. The views in this publication are those of Barclays Capital and are subject to change, and Barclays Capital has no
obligation to update its opinions or the information in this publication. Barclays Capital and its affiliates and their respective officers, directors, partners
and employees, including persons involved in the preparation or issuance of this document, may from time to time act as manager, co-manager or
underwriter of a public offering or otherwise, in the capacity of principal or agent, deal in, hold or act as market-makers or advisors, brokers or
commercial and/or investment bankers in relation to the securities or related derivatives which are the subject of this publication.
The analyst recommendations in this report reflect solely and exclusively those of the author(s), and such opinions were prepared independently of any
other interests, including those of Barclays Capital and/or its affiliates.
Neither Barclays Capital, nor any affiliate, nor any of their respective officers, directors, partners, or employees accepts any liability whatsoever for any
direct or consequential loss arising from any use of this publication or its contents. The securities discussed in this publication may not be suitable for all
investors. Barclays Capital recommends that investors independently evaluate each issuer, security or instrument discussed in this publication and
consult any independent advisors they believe necessary. The value of and income from any investment may fluctuate from day to day as a result of
changes in relevant economic markets (including changes in market liquidity). The information in this publication is not intended to predict actual results,
which may differ substantially from those reflected. Past performance is not necessarily indicative of future results.
This communication is being made available in the UK and Europe to persons who are investment professionals as that term is defined in Article 19 of the
Financial Services and Markets Act 2000 (Financial Promotion Order) 2005. It is directed at, and therefore should only be relied upon by, persons who
have professional experience in matters relating to investments. The investments to which it relates are available only to such persons and will be entered
into only with such persons. Barclays Capital is authorized and regulated by the Financial Services Authority ('FSA') and member of the London Stock
Exchange.
Barclays Capital Inc., US registered broker/dealer and member of FINRA (www.finra.org), is distributing this material in the United States and, in
connection therewith accepts responsibility for its contents. Any U.S. person wishing to effect a transaction in any security discussed herein should do so
only by contacting a representative of Barclays Capital Inc. in the U.S. at 745 Seventh Avenue, New York, New York 10019.
Subject to the conditions of this publication as set out above, ABSA CAPITAL, the Investment Banking Division of ABSA Bank Limited, an authorised
financial services provider (Registration No.: 1986/004794/06), is distributing this material in South Africa. Any South African person or entity wishing to
effect a transaction in any security discussed herein should do so only by contacting a representative of ABSA Capital in South Africa, 15 ALICE LANE,
SANDTON, JOHANNESBURG, GAUTENG, 2196. ABSA CAPITAL IS AN AFFILIATE OF BARCLAYS CAPITAL.
Non-U.S. persons should contact and execute transactions through a Barclays Bank PLC branch or affiliate in their home jurisdiction unless local
regulations permit otherwise.
In Japan, foreign exchange research reports are prepared and distributed by Barclays Bank PLC Tokyo Branch. Other research reports are distributed to
institutional investors in Japan by Barclays Capital Japan Limited. Barclays Capital Japan Limited is a joint-stock company incorporated in Japan with
registered office of 2-2-2, Otemachi, Chiyoda-ku, Tokyo 100-0004, Japan. It is a subsidiary of Barclays Bank PLC and a registered financial instruments
firm regulated by the Financial Services Agency of Japan. Registered Number: Kanto Zaimukyokucho (kinsho) No. 143.
Barclays Bank PLC Frankfurt Branch is distributing this material in Germany under the supervision of Bundesanstalt fuer Finanzdienstleistungsaufsicht
(BaFin). This material is distributed in Malaysia by Barclays Capital Markets Malaysia Sdn Bhd.
IRS Circular 230 Prepared Materials Disclaimer: Barclays Capital and its affiliates do not provide tax advice and nothing contained herein should be
construed to be tax advice. Please be advised that any discussion of U.S. tax matters contained herein (including any attachments) (i) is not intended or
written to be used, and cannot be used, by you for the purpose of avoiding U.S. tax-related penalties; and (ii) was written to support the promotion or
marketing of the transactions or other matters addressed herein. Accordingly, you should seek advice based on your particular circumstances from an
independent tax advisor.
© Copyright Barclays Bank PLC (2009). All rights reserved. No part of this publication may be reproduced in any manner without the prior written
permission of Barclays Capital or any of its affiliates. Barclays Bank PLC is registered in England No. 1026167. Registered office 1 Churchill Place, London,
E14 5HP. Additional information regarding this publication will be furnished upon request.