CMBS

SECURITIZATION RESEARCH
December 2009
U.S. SECURITIZED PRODUCTS

OUTLOOK 2010
BACK FROM THE BRINK
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
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.
Opposing forces: Prepayments in 2010 21

Credit and underwriting should continue to be the dominant factors driving prepayments in
2010. Over the next year, delinquency buyouts will be a major driver for both conventional
and GNMA speeds. However, the different approaches to underwriting standards (between
FHA and Fannie-Freddie) should lead to distinctive dynamics in their prepayments.
Top trades for 2010 29

We highlight key themes in agency MBS that should drive relative in 2010. In each case, we
consider specific trades to express these themes.
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
Agency underwriting: When will the squeeze ease? 96

We look at the underlying causes of current tight credit in the mortgage market. Risk
aversion, loan repurchases, economic uncertainty, and regulatory scrutiny will gradually
give way to the need to build origination volumes and increase profits. In particular, in H2
10, there could be a meaningful expansion of conventional credit to currently underserved
segments of the market.
Sub-prime speeds: Assessing credit curing 106

Voluntary prepays have ground to a halt even for low LTV sub-prime borrowers who have
never been delinquent. But an eventual easing of credit conditions and possible resumption
of new issuance could significantly impact sub-prime valuations if prepay speeds start to
edge higher. We discuss the impact of such credit curing on various sub-prime vintages.
21 December 2009 3
OVERVIEW
Back from the brink

Ajay Rajadhyaksha US securitized markets should pose far less systemic risk in 2010. As the strong
+1 212 412 7669 correlation between asset classes that characterized 2009 draws to an end, investors
ajay.rajadhyaksha@barcap.com are shifting their focus to relative value opportunities.
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.
Stepping back from the limelight

US securitized markets pose far It is time for US securitized markets to stop hogging the limelight. For several quarters,
less systemic risk for the system investors have pored through each delinquency data-point anxiously, policy makers have
conducted stress tests on asset-backed loans of all types, and banks have trembled at the
losses from these assets. 2010 should be the year when securitized markets return to the
shadows, at least in a systemic sense.
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.
Rates markets and the Fed: A benign macro picture

Rates should rise in 2010, but An overview of securitized assets always depends on the fixed income backdrop. We start
not sharply enough to hurt by noting that rates should rise in 2010. One reason is supply-demand imbalances. The US
securitized markets Treasury will continue to term out debt, which means Treasury coupon issuance in 2010
should be greater than in 2009, even with a smaller deficit. In fact, as Figure 1 shows, net
fixed income issuance in 2010 should be as big as in 2009.1 Meanwhile, there is a big hole
on the demand side. We expect the Fed to walk away from asset purchases after next
March, and with it will go the $1.7 trillion in duration demand that helped yields. Compare
2010 supply with 2009 numbers after excluding Fed buying (see the 2009 Ex-Fed column in
Figure 1): we are looking at a big jump in net supply in 2010. There are some offsets. Rising
demand from banks, US households and foreign investors should help fill the Fed gap. And
our economists expect inflation to stay mild, which could help yields. That is why we are
calling for a relatively mild sell-off (Figure 2); the forecast shows the 10-year Treasury rising
only to 4.5% – a 100bp move over 12 months. This kind of rate move should not unduly
distress securitized investors.
1
We ignore T-bill comparisons between 2009 and 2010 in these numbers.
21 December 2009 4
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
Source: US Treasury, Flow of Funds, Barclays Capital Source: Barclays Capital
21 December 2009 5
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
Figure 5: Foreclosures have not led to housing supply Figure 6: HAMP is not a silver bullet
Homes (000s) Cum Delinquency

3,000 100%
2,500 80%
2,000 60%
1,500 40%
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
0-10 10-20 20-30

Real-estate Owned Foreclosure
30-40 40-50
Source: Loan Performance, Barclays Capital Source: Loan Performance, Barclays Capital
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 $5.3. As
every month passes without a sharp increase in the REO bucket or a sharp drop in home
prices, the tail risk posed by housing declines ever so further.
Tail risk is also reducing in non-agency MBS. For the first time in a long while, there are
signs of credit burn-out. Current to delinquent roll rates have stabilized in recent months in
sub-prime, partly owing to burnout. There is some good news on severities as well; some
servicers are turning to short sales to avoid liquidation costs and recent severity data show a
down-trend in most agency sectors. Admittedly, while tail risk is receding over time, our
base case loss estimates are not much better than earlier in 2009. There will be more losses
in non-agency MBS – and lots of them.
Figure 7: Bank Losses over the next few years ($bn)

Assets ($bn) Proj. Cum Loss(%) Loss 2010 2011 2012
1st lien mortgages 1741 11% 200 75 63 63

2nd lien mortgages & HELOCs 855 21% 178 79 49 49
Multifamily Residential 216 6% 13 4 4 4
CRE mortgages 1090 9% 101 34 34 34
Construction loans 492 40% 195 98 98 0
C&I 1276 7% 85 28 28 28
Cards 393 13% 52 22 15 15
Others 1352 4% 60 20 20 20
Total 7415
Source: FDIC, Barclays Capital
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
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.
Problem areas: CMBS and tight residential mortgage credit

Stabilization in CRE lags an One big area of bank losses is the commercial mortgage/construction loan sector.
economic recovery Securitization rates for commercial mortgages have been much lower (25-30%) than for
residential loans (around 80% for sub-prime). As a result, banks have very significant exposure
to commercial real estate (CRE) loans. Indeed, the Fed seems worried enough that it has
extended TALF for CMBS into 2010. Moreover, CRE always lags an economic recovery. As
Figure 8 shows, we expect demand for CRE to stop shrinking only in the second half of 2010.
Investors should prepare for considerable stress and losses in CRE in 2010, but against the
backdrop of an economic recovery. Ironically, CMBS has been one of the best performing
sectors in the Barclays Aggregate index this year, and we still see value at the top of the capital
structure. But that is not due to fundamentals (for our detailed views on CMBS, see “Mind the
Gaps” on page 61). Instead, as with other sectors, the tail risk in CMBS (where many bonds
were priced to depression-era scenarios) has dissipated. While that might not seem like much,
it is the best news CRE might have for the next few quarters.
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
Sq ft (millions) YOY % Chg in NFP 770

300 forecast 4% 760
3%
200 750
2%
100 740
1%
0 0% 730
OFF
-100 HOT -1%
RET 720
IND -2%
-200
APT -3% 710
YOY % Chg in Payrolls (rhs)
-300
-4% 700
-400 -5% Jan 05 Dec 05 Dec 06 Nov 07 Nov 08 Nov 09
90 91 93 94 95 96 98 99 00 01 03 04 05 06 08 09 10 11 FH 30y FN 30y
Note: Shaded areas are recessionary periods. Source: PPR, BLS Source: Fannie Mae, Freddie Mac, Barclays Capital
21 December 2009 8
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.
Lower systemic risk – Making money through relative value in 2010

Spread products should be
helped by heavy sovereign debt After discussing the various securitized asset classes, the one theme that stands out is the
issuance… lessening of systemic risk. The main reason for this has been the monumental efforts of the
US government, including Treasury and the Fed. The other is simply the passage of time,
which has allowed banks to recapitalize themselves and the economy to stabilize, all helped
by low rates. The consequence of lower risk is that unrelated asset classes should no longer
move in the same direction. This was definitely true in 2009; as investors panicked,
correlations for completely unrelated assets jumped. For example, oil, equities, emerging
market debt and securitized assets all collapsed together in late 2008 and early 2009. And
from Q2 2009, liquidity pumped by the Fed and the recovery trade pushed up every asset
class. 2010 should be different, forcing investors to look for relative value.
…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.
Things to watch out for, and risks to beware of

Even with our benign view of spread products, there are some risks we are concerned
about. One is regulatory and accounting changes. H.R. 4173, the Wall Street Reform and
Consumer Protection Act of 2009, is working its way through Congress right now.
Meanwhile, SFAS 166/167 will go into effect next year. The immediate impact is probably
related to whether securitized trusts need to be consolidated on balance sheets or not. The
21 December 2009 9
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.
It is easier to comment on liquidity requirements. Regulators worldwide will force banks to

have bigger liquidity portfolios; for example, the FSA in the UK now requires British banks to be
able to withstand three months of market stress. Unlike in the UK, details in the US and
elsewhere are yet to come out. But an analysis of the FSA rules suggests that banks might
have to sharply increase their holdings of liquid and high-grade assets, though the funding
mix will play a big role (the lower the wholesale funding, the lower the liquidity portfolio
needed). The UK does not include agency MBS or debt as liquidity instruments, but we expect
US banking regulators to do so. There will probably be a phase-in period (it is four years in the
UK). But as and when the US rules become clearer, banks should start to position for the new
liquidity regime. One result could be a stronger bid for agency MBS (“Overview,” Market
Strategy Americas, October 22, 2009), but we expect this only from 2011.
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
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
AGENCY MBS
Mortgage basis outlook: Life after Fed

Kumar Velayudham The Fed’s $1.25trn purchase program had an outsized effect on the agency MBS
+1 212 412 2099 basis in 2009. Here, we examine how the mortgage market will fill the void when the
savelayu@barcap.com purchases stop after March.
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
Figure 3: Mortgage current coupon OAS (bp) Figure 4: Sector excess performance versus Treasuries (bp)
200 2006 2007 2008 2009 YTD

US Agg 85 -206 -710 713
150 MBS 122 -177 -232 532
Agencies 75 -52 -110 226
100 ABS 87 -634 -2223 2357
CMBS 137 -532 -3274 2710
50 Corporate 126 -523 -1988 2072
-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
Review of 2009: Basis tightening and index outperformance

As we began to develop our longer-term outlook over the past few months,2 it was clear
that the Fed’s role in the market was overwhelming in 2009 and that its eventual exit could
be similarly important. The Fed has thus far purchased $1.086trn in agency MBS and is set
to add approximately $165bn more through Q1 10 (Figure 1).
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.
As a consequence of this supply-demand imbalance, the mortgage basis tightened

considerably throughout 2009 (Figure 3). After widening to over L+100 in late 2008, the
current coupon has now rallied to levels that appear rich from a historical perspective. But
how did agency MBS fare against other spread product classes? Interestingly, agency MBS
Despite the strong performance, lagged the performance of mortgage credit products (Figure 4). For example, while agency
agency MBS lagged the MBS outperformed Treasuries by more than 500bp on a duration-adjusted basis, this was
performance of mortgage less than one-fourth of the excess performance of the ABS and CMBS sectors.
credit products
Significant secondary effects of Fed purchases
The Fed’s actions have had significant effects on MBS beyond the basis tightening. These
spill-over effects have influenced day-to-day trading, hedging activity, and funding levels for
agency MBS.
Tradable float has been significantly reduced

One by-product of purchasing $1.25trn in agency MBS out of a $5.2trn market is that
liquidity in many sub-sectors can be greatly reduced. Figure 5 shows the outstanding
balances of conventional (Fannie and Freddie) 30y fixed rate MBS by coupon. From these
balances, we remove bonds purchased by the Fed and pools pledged to CMO trusts. This
leaves only 50% of the pools available to be traded (Figure 5). And if we consider the GSEs’
retained portfolios or other MBS owned in held-to-maturity (HTM) accounts, the tradable
float starts to look even thinner.
2
Please see The day the Fed stood still, October 9, 2009.
21 December 2009 13
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
Implied funding for mortgages is at historical lows

Dollar rolls and implied financing rates go hand in hand with the reduced float. At the end of
2008, funding for agency MBS was noticeably more expensive than 1m Libor (Figure 6). A
by-product of the Fed’s relentless buying (especially during the first few months of the
program) was that implied financing rates in the dollar roll market declined sharply. This
effect was so powerful that rolls consistently traded to fail, with negative implied financing
rates, for much of the summer and fall. As a result, toward the end of 2009 the Fed sold
rolls in 5s and 5.5s to keep a technical situation from developing.
Mortgages traded to short empirical durations in 2009

Empirical durations were With Fed holding mortgages prices relatively steady, empirical durations were much shorter
shorter than model, despite than model effective durations during the year (Figure 7). As a result, MBS seemed to
higher callability of models outperform in a sell-off and underperform in a rally when hedged to model durations. Some
of the key benefits of shorter empirical durations have been better hedged carry and lower
mortgage volatility. Figure 8 shows that without the secondary effects of Fed purchases
(richer rolls and lower empirical durations), hedge-adjusted mortgage carry would have
been significantly lower.
Figure 7: Mortgages have traded to short empirical durations Figure 8: Limited carry without Fed support (32s/month)
10 4.5s 5.0s 5.5s 6.0s

TBA Carry 11 11.5 10.75 10
8 10Y Carry 14 14 14 14
Empirical HR 0.47 0.35 0.24 0.17
6 Model HR 0.67 0.52 0.39 0.34
4 Empirical Hedged Carry 4.4 6.6 7.4 7.6

Model Hedged Carry 1.7 4.2 5.4 5.2
2
Roll Richness 1.7 2.1 2.5 1.8

0
Dec-06 Jun-07 Dec-07 Jun-08 Dec-08 Jun-09
Carry Without Fed Impact 0.0 2.1 2.9 3.4
Empirical Duration Model Duration
Source: Barclays Capital Source: Barclays Capital
21 December 2009 14
The Fed and agency MBS in 2010

Now that we have considered the Fed’s role in agency MBS in 2009, we turn our attention
to 2010. We believe it is highly unlikely that the Fed will increase the size of the purchase
program beyond $1.25trn or extend the timeline for purchases. Aside from further reducing
the float and liquidity of the agency MBS market, additional purchases could face
congressional opposition. Moreover, assuming that the economy grows at 5% in Q1 10, as
forecast by our economics team, further quantitative easing (QE) is unlikely.
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.
Supply: Few changes from 2009

We expect about $10bn in With the Fed out of the picture, we examine the overall supply and demand landscape for
monthly run-off each from the agency MBS in 2010. Net supply should remain at $30-35bn per month, consistent with the
Fed and GSE portfolios run-rate of the past two years (Figure 9). Net supply will likely be driven primarily by new
and distressed home sales. Adding to this is the incremental run-off from the Fed and GSEs’
portfolios. Although we expect a subdued prepayment environment, the Fed also owns
higher-coupon MBS that could experience meaningful speeds. Figure 10 shows the
distribution of the Fed’s MBS holdings – note the sizeable amount in >=5% coupons. We
expect about $10bn in monthly run-off from the Fed, or $125bn for the year.
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)
Net issuance ($bn) 450

100 400
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 ($bn)
Source: Fannie Mae, Freddie Mac, Ginnie Mae, Barclays Capital Source: Federal Reserve
21 December 2009 15
Figure 11: Composition of GSE retained portfolio ($bn) Figure 12: GSE delinquency pipeline (%)
800 Serious delinquency rate

5%
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.
Demand side: The usual suspects

Over the last five recessions, the We now examine the demand side to see whether private demand can accommodate the
banking system increased its $625bn in agency MBS supply in 2010. The biggest sources of demand should be the usual
security holdings by 15% in the suspects – banks, money managers, and foreign investors.
year following the recession
Banks: A steep curve and a cash surplus should keep demand strong
Banks face an important decision in 2010 – should they expand their loan books or their
securities portfolios? To answer this question, we look at some of the factors that drive the
choice:
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
Figure 13: Bank securities versus loans, 1y change (%) Figure 14: Bank MBS holding versus curve
25% 3 60,000
20% 2.5 50,000
15% 2 40,000
10% 1.5 30,000
5% 1 20,000
0% 0.5 10,000
-5% 0 -
-10% -0.5 (10,000)

1970 Q1 1977 Q1 1984 Q1 1991 Q1 1998 Q1 2005 Q1 Mar-95 Mar-98 Mar-01 Mar-04 Mar-07
Recession Securities Loans Agency MBS Addition ($mn, RHS) 1s5s slope (%)
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.
Foreign demand: Risk reversal

During the credit crisis of 2008-09, foreign investors allocated away from riskier US assets
(agency debt, MBS, and corporates) and into Treasuries (Figure 17). Specifically, foreign
holdings of bills surged by $400bn. This risk aversion seems to have normalized over recent
months – MBS holdings have stopped falling, and foreign investors have continued to
reinvest paydowns. We expect a continued, yet moderate, risk reversal. We expect foreign
reserves to grow by $1trn in 2010, of which $400-500bn should find its way back to US
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
Source: FDIC, Barclays Capital Source: FDIC, Barclays Capital
21 December 2009 17
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
Long-term securities 6,262 7,162 9,136 9,463 8,792 8,212 9,528

Equities 2,144 2,430 3,130 2,969 2,101 1,867 2,528
Debt 4,118 4,733 6,007 6,494 6,691 6,345 7,000
US Treasury 1,599 1,727 1,965 2,211 2,634 2,439 2,733
Agency MBS 264 386 570 773 645 599 600
Agency debt 527 599 735 691 651 640 588
Corporate 1,729 2,021 2,738 2,820 2,761 2,667 3,079
Short-term debt 602 615 635 858 1,158 1,239 1,223
Total 6,864 7,778 9,772 10,322 9,950 9,451 10,751
Source: TIC, Barclays Capital
Money managers: Entry at wider spreads

Money manager universe is now When the Fed purchases end, monthly supply could spike to $50-60bn, helped by Fed and
significantly underweight agency GSE pay-downs. Strong demand from banks and foreign investors may not be able to
MBS, due to tight spreads absorb this supply; however, we expect that money managers will also add MBS in size and
help keep spreads contained. As stated earlier, we believe that the money manager universe
is now significantly underweight agency MBS, due to tight spreads. Hence, money
managers are unlikely to buy until spreads widen. The question becomes, “at what spread
level do MBS look attractive to money managers?”
Figure 18: Supply vs demand, post-Fed purchases (monthly) Figure 19: Money manager MBS allocation
Supply % in Agency MBS

Net
Net Issuance $30-35Bn Assets
Fund Name ($bn) 12/17/09 12/31/08 Change
Fed PayDowns $10Bn
PIMCO Total Return Fund $185.6 22% 62% -40%
GSE Paydowns $10-15Bn Vanguard Total bond Market $67.7 36% 37% 0%
Total $50-60Bn Index
The Bond Fund of America $37.9 17% 15% 2%
Demand Dodge & Cox Income $17.9 41% 47% -5%
Banks $15-20Bn Fidelity Advisor Total Bond $12.0 12% 14% -1%
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%
Combined $344.0 -23%
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
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.
Basis call: A stable basis at wider levels

We anticipate that spreads could We expect spreads to remain choppy at the start of the new year, with the Fed still a force to
widen 30-40bp before money reckon with. As we get closer to 2Q10 and the Fed exit nears, spreads should widen steadily.
managers come back and Bank and international demand may not be strong enough to hold the basis at these rich
stabilize the basis levels. These investors might also be wary of the impact of the Fed’s exit and might wait for
wider spreads before adding MBS in size. Consequently, we anticipate that spreads could
widen 30-40bp before money managers, who are currently underweight the index, come
back and stabilize the basis.
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.
Things to watch out for

While our base case is for relatively stable spread widening, convexity flows could change that
story. If rates rise sharply, MBS could extend and trade to much longer durations. Tight
underwriting and low HPA also add some extension risk. In a sell-off, convexity flows from
servicers could exaggerate spread widening. Our rates forecast, however, calls for a slow rise
in rates in the first half of the year, so a big convexity-shedding episode is not our base case.
21 December 2009 19
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
AGENCY MBS
Opposing forces: Prepayments in 2010

Derek Chen We expect credit and underwriting to continue to drive prepayments in 2010.
+1 212 412 2857 Delinquency buyouts are also set to be a major driver of conventional and GNMA
derek.chen@barcap.com speeds. However, the different approaches to underwriting standards between FHA
and Fannie-Freddie should lead to distinctive prepayment dynamics.
Wei-Ang Lee
Conventional voluntary prepayments have probably bottomed. The risk now lies to
+1 212 412 5356
the upside, although refinancing sensitivity to rates is unlikely to improve until at
weiang.lee@barcap.com
least H2 10.
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.
For GNMA, continued tightening in underwriting by the FHA should reduce

callability and significantly compress the medium-term GNMA/FNMA speed
differential for premiums.
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.
A slow upward trend for voluntary prepays

In 2009, there was an epic slowdown in prepayments. Adjusted for the level of mortgage
rates, speeds dropped far below historical norms, as extremely tight underwriting, the
decline in home prices, and increased refinancing costs offset nearly 100bp of refinancing
incentive (Figure 1). At these levels, there is little room for prepays to slow further, and the
risk now lies to the upside.
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
50 Dec 2009 report

30
40
20
30
20 10
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
Ultimately, the return of refinance-ability depends on the recovery of HPA, normalization of

underwriting, and reduction in risk premiums. When that happens, not only will overall
speeds rise, but the dynamics among different products will also change. For example, we
would expect credit-impaired loans such as high-LTV, low-FICO, high-SATO, and 10/20 IO
pools all to turn faster than average collateral when in the money (ITM), as they did during
2005-07. But such a scenario is unlikely to happen next year. For 2010, HPA should remain
weak, delinquencies will likely push higher, and banks should experience record levels of
loan losses and indemnification to the GSEs. Against this backdrop, it is difficult to imagine
lenders dramatically loosening underwriting standards any time soon. Consequently,
underwriting should remain very tight for H1 10. After that, there could be a moderate
downward drift in credit. Origination would continue to concentrate in low LTV loans, as weak
housing and high rescission rates by PMI companies would diminish lender appetite for
greater-than-80% LTV loans. However, there could be a downward drift in credit score, as last
year’s tightening in underwriting has pushed origination FICO far above historical norms.
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.
Buyouts to surge, causing significant tiering in speeds

While voluntary prepays should remain tepid during 2010, delinquency buyouts are likely to
surge, causing significant tiering among different products, vintages, and coupons. For the
past two years, FNMA has let delinquencies accumulate, leading to a seven-fold increase in
Figure 3: Opposing forces: Prepayments in 2010 Figure 4: Timeline of delinquency buyouts (FNMA 2006 6s)
CPR Involuntary prepayments (CPR) Modification CDR

50 12 Foreclosure CDR
Total CDR
40 10
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
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
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.
Although FNMA is required to repurchase a loan once it becomes 24 months delinquent,

loans in the foreclosure process are exempt from this restriction. Because most loans
should have begun the foreclosure process before the 24-month limit, such a threshold
should not be a dominant driver of FNMA buyouts.
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
Figure 6: FHLMC/FNMA 1y prepayment projections

FHLMC 30y Fixed FNMA 30y Fixed
Total CPR by no-point 30y mortgage rate Total CPR by no-point 30y mortgage rate
Vintage Coupon CDR 4.50 4.75 5.00 5.25 5.75 6.25 CDR 4.50 4.75 5.00 5.25 5.75 6.25
2003 5.0 0.8 32.3 22.5 17.6 11.7 7.5 6.0 1.0 30.7 21.2 16.5 10.7 6.9 5.7
5.5 1.2 35.9 27.5 22.0 15.4 10.2 8.2 1.5 33.0 24.8 19.7 13.4 8.8 7.6
6.0 2.1 27.9 23.1 19.4 14.9 11.6 9.8 2.6 24.5 20.2 16.8 12.8 10.2 9.2
2004 5.0 1.5 32.9 23.1 18.2 12.1 7.7 6.3 1.8 32.9 23.3 18.3 12.3 8.0 6.6
5.5 2.2 33.6 26.4 21.5 15.5 10.2 8.2 2.7 31.6 24.7 20.1 14.4 9.6 8.0
6.0 3.3 31.4 26.5 22.5 17.7 12.2 10.1 4.0 28.8 24.4 20.7 16.3 11.4 9.9
2005 5.0 3.1 35.4 26.7 21.4 14.9 9.6 7.9 3.8 33.7 25.3 20.2 13.9 9.2 7.9
5.5 4.7 33.2 27.4 23.0 17.8 12.3 10.1 5.8 31.3 25.9 21.9 16.9 12.1 10.4
6.0 7.2 30.4 27.1 24.0 20.3 15.3 13.0 9.0 27.1 24.4 21.8 18.7 14.7 13.3
2006 5.0 3.4 41.5 32.2 25.7 17.8 10.8 8.6 4.2 41.6 32.5 26.0 18.2 11.3 9.2
5.5 5.0 47.0 40.5 34.1 26.2 16.1 12.0 6.2 45.1 39.1 32.9 25.4 16.0 12.4
6.0 6.4 41.3 37.7 33.2 27.7 18.6 14.1 7.9 38.1 35.1 31.1 26.1 18.0 14.3
6.5 10.4 35.8 33.9 31.3 28.1 23.5 21.1 13.0 32.4 31.2 29.2 26.7 23.2 22.1
7.0 15.4 38.6 36.8 34.5 31.6 27.4 25.2 19.1 36.9 35.8 34.0 31.7 28.5 27.5
2007 5.0 3.3 40.4 30.7 24.3 16.4 9.9 8.0 4.2 40.9 31.2 24.8 17.0 10.6 8.8
5.5 4.8 43.8 37.3 31.2 23.8 14.8 11.2 6.0 44.1 37.8 31.7 24.3 15.6 12.1
6.0 8.0 39.9 36.5 32.5 27.6 19.4 15.1 9.9 38.8 35.8 32.1 27.5 20.0 16.2
6.5 13.2 36.5 34.9 32.9 30.3 26.3 23.8 15.7 34.6 33.6 31.9 29.9 26.7 25.0
7.0 22.7 46.1 44.6 42.5 39.9 35.9 33.3 25.0 44.0 43.0 41.3 39.2 36.1 34.4
2008 5.0 1.0 38.6 27.4 20.9 13.0 7.3 5.7 1.2 36.9 26.0 19.8 12.2 6.8 5.5
5.5 2.4 40.7 33.1 26.9 19.4 11.4 8.5 2.9 38.8 31.5 25.6 18.4 10.9 8.4
6.0 4.4 38.5 34.7 30.5 25.3 16.8 12.2 5.5 36.5 33.1 29.1 24.2 16.3 12.4
6.5 8.8 33.6 31.8 29.6 26.9 23.1 20.7 11.0 30.1 29.1 27.4 25.3 22.5 21.3
7.0 15.1 40.2 38.4 36.2 33.4 29.5 27.2 17.3 37.3 36.2 34.4 32.2 29.4 28.1
2009 4.5 0.1 25.3 11.5 4.9 3.6 3.1 3.1 0.1 25.3 11.5 4.9 3.6 3.1 3.1
5.0 0.3 36.9 26.0 18.2 9.3 3.8 3.1 0.3 36.9 26.0 18.2 9.3 3.8 3.1
Source: Barclays Capital
GNMA/FNMA speed differential to compress

While conventional prepayments are likely to trend higher in 2010, the roadmap for GNMA
looks different. Specifically, GNMA callability should reduce because FHA and GNMA have
started tightening underwriting. Of particular note are two recent changes made in late 2009:
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:
Increase the minimum FICO score
Increase the down-payment requirement
Increase the upfront and annual mortgage insurance premium (MIP)
Increase lender accountability and indemnification for wraps and warranties
Tighten the requirement for lender qualification, broker supervision, and appraisals
21 December 2009 25
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.
Turnover to slow to 4-5 CPR, GNMA to provide extension

protection
In a sharp sell-off, conventional With many investors calling for much higher mortgage rates in 2010, the MBS universe could
discount speeds should be subject to significant extension risk. Cash-out refinancing should remain nearly nonexistent
drop below home sales, for the next year, so discount speeds should collapse to housing turnover. Figure 9 shows
reaching only 4-5 CPR. percentage of existing home sales versus FNMA 30y out-of-the-money (OTM) speeds.
Against historical norms, discount speeds have dropped below existing home sales. This is
because a large portion of home sales are now distressed sales associated with borrower
defaults, which are less prevalent among conventional lower coupons.
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
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
75bp OTM Speeds (CPR) Cumulative % of FNMA 30y universe

% Existing Home Sale
11 100%
75bp OTM FNMA 30y CPR Jan-03
10
80% Jan-08
9 Jan-09
8 60% Nov-09
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
FICO % Cum Default

780 10%
06q1 07q1
760 08q1 08q2
8%
740 08q3 08q4
720 09q1
6%
700
680
4%
660
640 FHA
2%
620 FNMA
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
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
FHA CPR Average Life(yr)

100 11.8
FNMA 14 12
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
AGENCY MBS
Top trades for 2010

Matthew Seltzer We highlight a number of key themes that should drive relative value in the agency
+1 212 412 1537 mortgage market in 2010. In each case, we consider specific trades to express these themes.
matthew.seltzer@barcap.com
Theme 1: Fed exit from the market

Kumar Velayudham
+1 212 412 2099 The end of the Fed’s agency MBS purchase program will be an important event in 2010 and
savelayu@barcap.com have significant ramifications for relative value. In “Mortgage basis outlook: Life after the Fed”
on page 12, we noted that the basis is likely to be pressured in Q1 10. In addition to the
Nicholas Strand obvious supply and demand issues, there are two arguments in favor of a short basis position.
+1 (212) 412 2057 First, valuations in agency MBS are still quite rich, relative to Treasuries, Agency Debt and
nicholas.strand@barcap.com Corporate Credit. Second, MBS could trade to longer durations. Despite low speeds,
mortgages have traded to short empirical durations, as the Fed has kept a lid on any spread
moves. We think that this will change by Q2 and MBS price volatility could increase.
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).
Coupon swaps hedged with Treasuries – Similar to a long IO position, up-in-coupon

swaps can be a short basis trade, but with more liquidity. However, technical factors in
the coupon stack and rolls can present other difficulties. Right now, we believe the 6/5
swap makes the most sense.
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.
Mortgage options – Mortgage options allow investors to implement conditional spread

wideners. We recommend selling call options on MBS and hedging with receiver
swaptions on 2s and 10s.
21 December 2009 29
Figure 1: Summary of various short basis options

Positives Negatives
Short TBAs Liquidity Moderate negative carry

Ability to short specific coupon Daily hedging
IO hedged with TSY Significant carry Less than perfect correlation

Muted prepay Lower liquidity
Coupon swap hedged with TSY Significant carry Less than perfect correlation
Muted prepays Coupon swap technicals
CMM-CMS FRAs Clean execution, no hedging Lower liquidity

Positive carry Convexity related technicals
Mortgage options Conditional spread trades Short horizon

Ability to short specific coupon Lower liquidity
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.
Theme 2: Rolls to cheapen in 2010

The Fed’s involvement in 2009 has blessed the roll market with very low implied funding for
Rich dollar rolls made
much of the year (Figure 2), with rolls consistently trading through fail. Hence, selling the roll
specified pools less
was often more advantageous than owning specified pools outright. For example, in
attractive for much of 2009
Figure 3, we look at monthly carry from owning 2008 LLB FN 5.5s along with the monthly
drop for FN 5.5s since the start of the year. At the start of 2009, there was still a carry
advantage to owning specified pools. But as implied financing rates dropped, this changed.
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)
2.5 1m Libor Implied Roll Funding GC 14
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
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.
Figure 4: Relative value across specified pools

Rolls Normalizing in Q2 09 Rolls Normalizing in Q3 09
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)
5.0% TBA 103-25 19.0

2005 103-28 3 15% 18.1 -1.6 0.0
2004 104-00 7 22% 15.9 -0.6 0.5 183
LLB 104-06 13 17% 9.0 3.9 40 2.7 57
MLB 104-00 7 13% 10.2 3.8 22 2.7 32
HLB 103-29 4 15% 12.0 2.9 16 2.2 21
5.5% TBA 105-20 25.9
2006 105-22 2 12% 25.4 0.2 129 0.1 194
2005 105-26+ 6.5 14% 20.2 6.4 12 4.2 18
2004 105-30 10 18% 18.8 7.5 16 5.0 24
LLB 106-20 32 35% 13.5 11.1 34 7.4 52
MLB 106-10 22 36% 16.4 8.6 31 5.8 46
HLB 105-31 11 36% 20.6 4.9 27 3.3 40
6.0% TBA 106-23+ 31.0
2006 106-28+ 5 18% 26.4 7.9 8 5.3 11
LLB 108-03+ 44 49% 16.2 19.1 28 12.7 42
MLB 107-27+ 36 61% 19.6 13.9 31 9.3 47
HLB 107-05+ 14 50% 24.7 8.8 19 5.9 29
6.5% TBA 107-24 29.4
2006 107-24 0 0% 25.3 9.5 0 6.3 0
LLB 109-08 48 57% 16.9 17.8 32 11.9 49
MLB 108-28 36 69% 20.2 12.5 35 8.3 52
HLB 108-05 13 63% 25.0 6.7 23 4.5 35
Theme 3: Changing prepayment environment

Aside from an underweight on the basis, the changing prepay environment should play a big
role in 2010. In “Opposing forces: Prepayments in 2010” on page 21, we discuss why
voluntary speeds should remain relatively subdued next year while involuntary speeds should
trend higher for credit impaired collateral. In this section, we outline our views on conventional
and Ginnie Mae’s passthroughs and discuss relative value across coupons.
21 December 2009 31
Overweight premium 30y conventional pass-throughs

When looking at valuations across the conventional 30y coupon stack, there are three
factors to consider. First, voluntary prepayments are likely to remain relatively subdued.
Second, involuntary prepayments (owing to buyouts) are likely to accelerate in 2010, but
the timing is unclear. Third, our short basis position needs to be overlaid with our
prepayment-driven valuations. We modify our Barclays Capital prepayment/OAS model in
order to mimic three different prepayment scenarios and compare the relative performance
of coupons across these scenarios:
Base case – Production model with no modifications or changes.
Muted voluntary speeds – We reduce voluntary prepayments to reflect the current

experience of agency prepayments. However, we assume that this effect is phased-out
over the next two years as the housing market strengthens and underwriting criteria
gradually ease.
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
4 TBA -11.1 -10.7 -10.7

4.5 TBA -6.2 -3.4 -3.4
5 TBA -29.5 -16.1 -16.9
5.5 TBA -28.4 -5.0 -7.2
2007 -37.9 -10.0 -14.7
6 TBA -22.2 14.1 9.3
2007 -19.2 16.3 7.6
6.5 TBA 1.7 41.3 30.1
2007 9.1 46.3 27.9
7 TBA -20.4 25.9 5.4
2007 -3.5 26.6 -4.8
Note: As of December 17, 2009 close. Source: Barclays Capital
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
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.
Underweight higher coupon GN/FN swaps

Prepay changes also create opportunities between conventionals and Ginnie Mae coupons.
While the themes (slower voluntary speeds, faster involuntary speeds) are the same, there are
important differences that should lead Ginnie Mae speeds to be faster than conventionals in
2010. Voluntary speeds for Ginnie Mae collateral have been faster than conventionals in 2009.
Over the next year, they should decline somewhat, but are likely to remain more robust than
conventionals in the near term. Proposed underwriting changes to the FHA program4 should
reduce their prepayments; however, the extent of tightening as well as the exact timing of any
future changes is unclear. While Ginnie Mae voluntary speeds should drop at the margin, the
difference in involuntary prepayment behavior is likely to drive Ginnie Mae relative value.
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.
Figure 6: GN and FN delinquency pipelines and current-to-delinquent roll rates (%)

90-day+ delinquency (%) C-Q roll rate (%)
Cpn Vintage GN FN* Diff GN FN* Diff
5 2006 5.7 4.9 0.8 9.1 4.0 5.1

2007 8.0 4.9 3.1 13.1 4.0 9.1
5.5 2006 6.0 7.2 -1.2 9.4 5.8 3.6
2007 8.4 7.0 1.4 11.7 5.7 6.0
6 2006 7.7 9.2 -1.5 12.4 7.4 5.0
2007 11.4 10.8 0.6 16.4 8.6 7.7
6.5 2006 12.5 13.6 -1.1 16.9 10.7 6.2
2007 16.8 18.5 -1.7 21.8 14.4 7.4
7 2006 14.3 20.1 -5.8 20.5 15.5 5.0
2007 23.3 26.3 -3.0 26.3 23.8 2.5
Note: Fannie Mae numbers are estimates. Source: Fannie Mae, Freddie Mac, Ginnie Mae, Barclays Capital
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
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.
Theme 4: Protecting against extension risk

As we discuss in the “Mortgage basis outlook: Life after the Fed” on page 12, higher Treasury
rates and wider mortgage spreads could lead to significant extension risk next year. Given that
we expect the banking sector to be a major presence in 2010, protection against extension
risk should matter that much more.
Overweight GN/FN lower coupon swaps

Previously, we recommended selling higher coupon GN/FN swaps, owing to the prepayment
characteristics of Ginnie Mae pass-throughs. However, these elevated speeds can help
investors concerned about extension. In Figure 9, we look at the relative performance of 4-5%
Ginnie Mae and Fannie Mae pass-throughs in our OAS model. As before, we adjust our model
to reflect a reduced voluntary prepayment environment going forward. The GN pass-throughs
consistently pick up 10-25bp of Libor OAS, while exhibiting shorter average lives.
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
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.
Figure 11: Extension risk across short-duration alternatives

Base Case +200
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
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
Buy agency IOs as a hedge against higher rates

A final alternative to manage extension risk is to add agency IOs. This is an excellent way to
boost carry while also positioning for a higher rate environment. The flip side, of course, is
that they do poorly when rates rally. But overall, agency IOs can help smooth the return
profile of an existing MBS portfolio. For example, in Figure 12 we look at the total returns
(including coupon returns, not just price returns) for a pass-through and a pass-through
plus IO combination. While the one-year returns for the pass-through turn negative if rates
rise more than 100bp, adding the same notional amount of a 5% IO created a far more
stable portfolio. In summary, the IO fulfils two functions: it boosts base case yield for the
portfolio while protecting against rising rates.
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
1Yr Returns (%) CPR (%)

6 25% 50
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
Source: Barclays Capital Source: Fannie Mae, Barclays Capital
21 December 2009 36
RESIDENTIAL CREDIT
2009 a tough act to follow, but upside remains

Sandeep Bordia 2010 is unlikely to provide outsized return opportunities similar to February-March
+1 212 412 2099 2009, but there is still potential for meaningful upside.
sandeep.bordia@barcap.com
Non-agency unleveraged loss-adjusted yields are still significantly higher than those
of corporates, CMBS, and consumer ABS. Alt-A hybrid super seniors with low
Jasraj Vaidya
double-digit yields look attractive on an outright basis.
+212-412-2265
jasraj.vaidya@barcap.com Negative headlines on performance will likely persist, but we expect clear signs of
credit burnout in the performance of always-current loans in 2H10. This should
Sandipan Deb drive risk premiums lower, with option ARM super seniors and subprime last cash
+1 212 412 2099 flow floaters the biggest beneficiaries.
sandipan.deb@barcap.com
Cheap leverage from the repo market should continue to drive prices higher in early
Elena Warshawsky 2010, especially for higher dollar-priced securities. The risk of servicers recouping
+1 212 412 3661 advances post-massive modifications could be a problem for subprime front pays,
elena.warshawsky@barcap.com though.
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
$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.
Figure 1: Non-agency prices recover from March lows

Price Increase since
(%)
Price Dec-08 Mar-09 Jun-09 Dec-09 Mar ‘09 Dec ‘08
Jumbo Fix SS AAA 70 60 75 84 40 20

Jumbo Hyb SS AAA 57 45 60 73 62 28
Alt-A Fix SS AAA 58 43 56 71 65 22
Alt-A Hyb SS AAA 50 35 47 57 63 14
MTA SS AAA 44 33 40 48 45 9
ABX 06-1 PAAA 90 83 84 88 6 -2
ABX 06-1 AAA 80 61 69 80 31 0
ABX 07-1 PAAA 44 30 30 40 33 -9
ABX 07-1 AAA 40 25 26 33 32 -18
21 December 2009 38
Performance: Light at the end of the tunnel?

Current credit performance in the non-agency sector remains dismal despite the recent
slowdown in the pace of worsening. Close to half of all outstanding borrowers in 2007
subprime collateral (and 40% of 2007 option ARM) are classified as 60+days delinquent.
Voluntary prepayment levels for these borrowers are languishing in the low single digits.
Even in jumbo loans, the most pristine of non-agency collateral, more than 1% of all current
borrowers are turning delinquent every month.
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.
First signs of credit burnout?

Collateral characteristics could Credit burnout refers to the progressive improvement in the performance of loans as bad
improve because of burnout… borrowers exit the pool, resulting in a more favorable composition. To be sure, the
composition can improve or worsen depending on the environment. For example,
borrowers with better credit quality who are eligible for refinancing quickly prepay and exit
the pool as rates start to drop. On the other hand, loans with worse credit quality go
delinquent sooner and exit the pool through defaults. The past few years, however, have
seen limited refinancing opportunities for non-agency borrowers, especially alt-A and
subprime, and defaults have shot up, leading to positive composition shifts.
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
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
Another way to understand burnout is to look at the delinquency composition of subprime.

Of about 40-50% of borrowers reported as current in 2007 subprime, more than 70% have
not missed a single payment to date. They have managed this feat in arguably one of the
most trying environment in decades, and this should result in improved performance in the
back end. In 2010, we should see the beginning of some credit burnout effects, especially in
deals off older vintage subprime, option ARM, and worse-quality alt-A collateral, for which
default rates have been high. Recent months have already seen a leveling off in current-to-
delinquent rates in the subprime sector, and part of it can be attributed to burnout (and, of
course, better home price numbers over the past few months).
How will burnout affect future performance?

…leading to performance Any performance stabilization due to burnout will first be seen in loans that have always
stabilization in the longer term been current and unmodified – causing current-to-delinquent roll rates for these loans to
drop (Figure 5). However, this may not have an immediate effect on overall constant default
rates (CDRs) because of the large inventory of 60+day delinquent and modified loans that
are still in the system. In Figure 4, we construct a simple roll rate model to demonstrate the
effect of any potential credit burnout on overall subprime defaults under two scenarios. In
the first scenario, monthly current-to-delinquent transition rates remain at the current level
of around 6%. To show the effects of burnout, we model a second scenario in which
current-to-delinquent rates immediately fall to 3%.
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
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
HAMP redefault performance should be better than previous modifications

Earlier modification efforts have While credit burnout will have more of a long-term effect, the more immediate attention-
led to high redefault rates… grabbing metric will be re-defaults on HAMP modifications. There is not much HAMP re-
default data yet, but we can derive some broad conclusions using the performance of pre-
HAMP modifications.
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
Severities: Servicer behavior will remain important

Short sales have a positive Recent severity data have shown a downtrend in most non-agency sectors. While some of
effect on severities this can be attributed to favorable home price numbers, important shifts in servicer
behavior have also contributed to the decline. Servicers are increasingly going for short
sales to resolve delinquencies (Figure 8) as a way to avoid the high fixed costs incurred in
the foreclosure/REO process, including attorney fees, repair costs, and foreclosure fees.
Short sales also inherently imply fewer P&I advances. With the government moving to
encourage liquidations through short sales under the HAFA program, we expect their share
to continue to rise in the near future. Historically, short sale severities have been 15-20%
below the severities from real estate-owned (REO) liquidations (Figure 9). The increasing
share of short sales should, therefore, reduce overall severities for the deals.
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.
We expect timelines to continue to increase as the government pressures servicers to

expand the net of loans in trial modifications. However, if servicers stop advancing earlier,
the effect of extending timelines on severity will be muted. It may, at first glance, seem that
P&I advances do not change the total cash flows to the deal and that a change in headline
severity is compensated by more or less money paid to the deal earlier. While this is true,
the change in the timing of cash flows due to lower/higher advancing significantly alters
allocation across bonds and has important implications for valuations especially in
conjunction with mods, it could be almost a 5-10pt effect for some subprime front pays
(more on this in the valuation section).
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
Figure 8: Share of short sales by sector Figure 9: Severities by type of liquidation
Severity REO Shortsale

Jumbo Alt-A Negam Subprime 85
Share of shortsales (%)
55 75
50
45 65
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
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.
Prepayments: Refi eligibility will drive prime sector speeds

Prepays will continue to be slow While mortgage rates have touched all-time lows, there has simply been no refinancing
in subprime and option ARMs… market for loans falling outside agency criteria. Our outlook for prepays in 2009 depended
on refi eligibility into GSE/FHA loans alone. While that is still the primary driver for projecting
prepay speeds, some new clean jumbo origination in the non-agency space is an additional
factor to consider in 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
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
Government programs: Second innings

Fresh waves of government There is little doubt that strong policy action on slowing down foreclosure sales and on the
action expected in the secondary market helped reverse the price declines in non-agencies in 2009. However, we
next couple of years think that as the initial effect of many of the government programs wears off and their
shortcomings become more evident, we will see another stage in government intervention.
For example, HAMP is running into issues of too few permanent modifications, and re-
default performance is expected to be poor. At the same time, the number of homes in
foreclosure and deep delinquency continues to balloon. This leads us to believe that what
we have seen is only the first wave of government programs to tackle foreclosures. The next
year or two will likely see many changes and additions to these programs as the
government tries to keep foreclosures under tight control in the face of adverse
performance and elevated unemployment, especially as elections approach.
21 December 2009 44
Figure 13: Timeline of government programs and non-agency prices
Dollar Price Jumbo Fixed Alt-A Fixed Option ARM

100
90
80
70
60
50
40
30
Aug-08 Nov-08 Feb-09 May-09 Aug-09 Nov-09
H4H HAMP PPIP HAMP PPIP purchases

announced announced announced active commence
Despite initial success, HAMP has many issues

HAMP dogged by issues that will HAMP was the first successful modification program after the failure of other efforts such as
become more apparent in 2010 H4H and FHA Secure. The program had a slow start due to operational hurdles at servicers.
Nonetheless, it managed to reach its November goal of 500k trial modifications and
continues to modify more loans (Figure 14). Combined with other initiatives, it also slowed
REO supply to the market, although foreclosures have ballooned.
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.
Low conversion from trial to permanent modification

Conversion rate remains low due The success of the HAMP program to date has been in terms of the number of trial
to documentation and modifications initiated. However, the conversion rate to permanent modifications remains low.
affordability problems As of the November HAMP report, only 30k loans have been converted into permanent
modifications of a total of 728,000 active HAMP mods. This problem will be highlighted early
in 2010 as the first set of trial modifications roll off without getting converted to permanent
mods. Congressional testimonies by some servicers show that the reasons for the failure of
trial modifications are almost equally divided into three categories: inability to make all trial
payments, non-submission of required documents, and submission of incomplete or
erroneous documentation. Theoretical eligibility for HAMP remains high even after accounting
for income misstatement at the time of origination and a reduction in income since. However
the government may have trouble reaching its goal of 3-4mn modifications given the above
mentioned reasons, unless measures are taken to alleviate these problems.
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
because of failed modifications and short sales, the number of advances that servicers need
to make will also increase.
Figure 14: Number of trial modifications initiated (‘000s)
Number of trial mod initiated ('000)
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
Potential program and policy changes in 2010

As pressure increases on Intended or unintended, the HAMP program has been successful in slowing the rate at which
government, changes in policy foreclosures enter REO. However, most of the loans modified in 2009 are likely to re-default or
likely in 2010 fail to become permanent modifications in 2010. With unemployment expected to remain
high, widespread foreclosures will not be politically palatable and may jeopardize the nascent
recovery. As the pressure to do more to help borrowers grows, we believe that the
government will take action to address some of the issues we have discussed above and
potentially more.
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.
Forgiveness instead of forbearance: Debt forbearance can reduce payments for

borrowers but still does not help the negative equity situation. Debt forgiveness reduces
borrower LTV, adds an incentive to continue paying, and provides an outside chance of
prepayment, as refinancing opportunities may become available. The handling of
second liens, opposition from investors who would be hurt, and potential moral hazard
issues as borrowers flock to get their balances reduced might impose challenges, but
this option is likely to be brought up time and again.
21 December 2009 46
Streamlining documentation requirements: Current HAMP guidelines require a

significant number of documents in order to become eligible for a permanent
modification. The Conversion Initiative recently launched by the Treasury has made
some attempts to minimize documents and simplify the process. We expect more
changes in this area next year.
H4H version 4.0?

Another iteration of H4H H4H preceded HAMP but has been a failure so far since it has hardly ever been used.
is also likely Multiple issues continue to dog the program despite attempts to improve it. The main
problems are the eligibility guidelines and trouble with extinguishing second liens. The
eligibility criteria for H4H, especially those related to DTI, are stringent and result in low
eligibility. Because refinancing a loan under H4H requires the extinguishment of all other
liens, existing second liens pose a significant problem. Congress has allocated significant
funds to the program and has a clear stake in making it work. Relaxing the eligibility criteria,
along with a better plan to tackle second liens, could help revive H4H. The program is a net
positive for the non-agency market because the loans are refinanced by FHA after a debt
write-down that is much smaller than the losses under other alternatives. Given the efforts
and funds already invested in this program, it is likely that the government will make
another attempt to make it work. However, servicers are still likely to favor other programs
such as mods, which allow them to keep the servicing IO strip unless they are offered
incentives or allowed to service the new loan.
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.
A better second-lien program

A better second lien program At present, no successful program exists to tackle second-lien loans. Like HAMP,
would help first lien valuations participation in 2MP is voluntary, but no major participants have signed up for it, A large
proportion of second liens and HELOCs are owned by banks that also happen to be the
largest servicers; therefore, low participation does not seem surprising. This remains a
major impediment to the implementation of H4H or instituting debt forgiveness
modifications. To many, it is also a moral hazard issue, with the first lien taking a rate or
balance hit before the extinguishment of the second lien. In fact, the second lien is one of
the bigger beneficiaries of the first lien modification, as they can potentially get more cash
from the borrowers. While negative for banks, more second-lien modifications would help
reduce back-end DTI and improve first-lien performance.
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
Selling loans from deals below par

Selling loans out of deals may Servicers remain restricted in terms of the number of loss mitigation actions that can be
open up many other alternatives implemented on a loan within a deal. However, if individual loans can be sold from the deal
at below par, then the loan can be handled more effectively using a variety of tactics that are
otherwise not possible within a deal. One such tactic would be renting out the home to the
delinquent borrower on the lines of the GSE program. There has been some talk of allowing
these kinds of sales, and this is a possibility in 2010. This could increase the number of loan
workouts, although valuation of the loan may be tricky.
Expected outcome of policy changes

Likelihood of these changes can As mentioned above, the valuation effect of many of these changes could be positive or
push up uncertainty premium… negative depending on the individual action, as well as implementation details. However, we
believe that two overriding themes will dominate in 2010:
Bouts of modification/legislation-related uncertainty: The government will likely do

whatever it takes to keep more borrowers in their homes and contain the supply of distressed
assets. As more programs/changes to existing programs are announced, there will be bouts
of uncertainty, and any misdirected government action could have a strong negative effect on
valuations. That said, a lot of this uncertainty is already priced in, and barring the spectre of
cramdowns or something close, the market should be able to absorb it.
… 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.
“Wall of cash” should keep secondary market technicals strong

Return of leverage and less While apprehension related to more drastic government intervention in modifications and
distressed markets bode well for the consequent uncertainty about performance persists, the technical landscape looks fairly
prices in 2010 strong. This is driven, in part, by liquidity conditions, which are expected to remain strong, at
least in the first half of 2010. The return of leverage to secondary markets through the PPIP
and, more importantly, through third- party repo markets is likely to bolster demand.
Furthermore, the major technical factors that led to a deluge of supply into the secondary
markets in the early part of 2009 are mostly missing, with rating agency downgrades and
bank capital close to becoming non-issues at this point. The insurance companies’ new
model-based regulatory capital regime may lead to some supply, but will likely be
manageable. Finally, while we expect some new non-agency issuance in clean collateral,
new supply will, at best, be minimal. All of these factors bode well for prices in early 2010.
Who owns non-agency AAA MBS?

Before going into detail on supply-demand issues, it is worthwhile to identify the major
participants in the non-agency market. We focus largely on original AAAs, as they
constitute 70-75% of notional outstanding and more than 90% of market value. In 2009,
the outstanding balance of non-agency AAA securities fell by about $250bn, to $1.20trn. As
of our latest estimate, the biggest owners by notional balance are commercial
banks/savings institutions and credit unions (Figure 15), followed by the GSEs and FHLBs.
Money managers/pension funds and insurance companies also own significant portions of
the outstanding. Overall, the relative positions are similar to at the start of this year, as some
21 December 2009 48
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
Commercial banks/S&L/credit unions etc 435 388

GSEs/FHLBs 370 270
Money managers/pension funds 208 170
Insurance companies 148 132
Overseas 137 99
Hedge funds 50 75
Broker/dealers 65 60
REITs/others 14 17
Total 1452 1202
Source: Federal Reserve, FDIC, Intex, SNL Financial, Barclays Capital
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
Jumbo FIX 8.1% 7.7% 8.7% 55.6% 20.0%

ARM 13.1% 7.6% 7.4% 36.7% 35.1%
Alt-A FIX 3.9% 2.3% 2.9% 34.7% 56.2%
ARM 6.3% 3.9% 4.7% 36.7% 48.4%
Alt-B Agg 1.8% 2.9% 2.1% 13.7% 79.5%
Negam NEG 1.1% 3.9% 2.0% 34.5% 58.5%
Subprime Agg 7.9% 5.4% 5.6% 34.5% 46.6%
Source: Intex, Bloomberg, Barclays Capital
21 December 2009 49
Major investors less likely to trigger fire sales

…and major investors in strong Second, each of the major investor groups is in a much stronger capital position, and the
capital positions risk of fire sales is much diminished.
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
1 0.40% 0.85% 76.65 1 0.30% 0.65% 76.50

2 1.30% 2.95% 78.31 2 1.00% 1.50% 77.16
3 4.60% 7.30% 81.98 3 2.00% 3.25% 78.55
4 10.00% 16.50% 91.02 4 4.50% 7.25% 81.94
5 23.00% 26.50% 103.40 5 10.00% 20.00% 95.00
6 30.00% 6 30.00%
Source: NAIC, Barclays Capital
Insurance company regulatory capital needs could lead to some selling…

Figure 17 shows the NAIC capital charges and maximum price breakpoints for a bond
Insurance companies could be a whose intrinsic value was determined to be $76 by the PIMCO model. The intrinsic price is
risk with the new capital rules… an output from PIMCO’s probabilistic loss model for each security. The maximum price
breakpoint is calculated as the intrinsic value divided by one minus the midpoint loss for
each designation. Let us further assume that the amortized cost basis for the bond is $90.
21 December 2009 50
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.
…especially from P&C insurers

So assuming that the fair value of the bond is $55, P&C insurers would be forced to hold the
bond at fair value, since it is categorized as designation 5. The new designation is
recalculated at the fair value, which in this case would be 1 (since $55<$76.5 which is the
maximum price for designation 1). So the P&C insurer can hold the bond at $55 and hold
0.3% capital against it. As for life insurers, since the initial designation was 4, they can
continue to hold the bond at an amortized cost basis of $90 and hold a 10% RBC charge
(for designation 4). So assuming that both insurers are carrying the bond at its amortized
cost basis, the P&C insurer would need to mark down the bond by $35, whereas the life
insurer has to take only a $10 capital charge for the same bond.
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.
Still enough cash on the sidelines to provide support to prices

Liquidity is freely available and Over the past nine months, about $600bn of cash has left money market funds, potentially
should remain a big positive, at looking for higher yielding assets. A large fraction of this has probably reached bond funds,
least in 1H10… whose assets under management increased by about $500bn over the same period. At the
same time, money market assets remain at fairly elevated levels of about $3.3trn (Figure
18), which is at least $1trn or so higher than levels seen in the past few years. While part of
this cash will start to fade away as the Fed withdraws liquidity from the markets next year, it
could potentially continue to chase yields on risky assets, including non-agencies. We
expect this to remain a strong positive for non-agency securities, since their unleveraged
loss-adjusted yields are still much higher than for other asset classes.
21 December 2009 51
Figure 18: Cash still waiting on the sidelines, $bn
4,000
3,500
3,000
2,500
2,000
1,500
1,000
MMKT funds Long Term Bond Funds
Source: Haver, ICI, Barclays Capital
The return of leverage through third-party repo

… which will have a noticeable There are a couple of ways this cash can make its presence felt in the non-agency markets.
effect on the improvement in While distressed funds can no longer meet their targets with the unleveraged yields
leverage availability and cost available in the market except for a few select areas, there is still a lot of real money
demand, as 8-10% yields still look substantially better than unleveraged yields in other
sectors. However, the easier transition for money market investors would be to provide repo
leverage to buyers with more risk appetite. We have already seen a sharp rise in availability
of third-party repo funding over the last 6 months of the year, with haircuts of 15-50%
across various products (Figure 19). While there are obvious risks to using third-party repo,
such as mark-to-market volatility and the short-term (typically 1-3 months) recourse nature
of the funding, if the leverage remains available even at current levels, it produces fairly
attractive yields (Figure 19). If there are signs that some of the uncertainty about
performance and modifications is subsiding, we will likely see a further compression in
haircuts and funding spreads. That would reduce the gap between unleveraged yields in the
sector and for other risky assets such as corporates. This form of leverage favors high-
coupon current-pay bonds. This is amplified by the fact that higher coupon jumbo bonds
with higher prepays are also the bonds for which the largest amount of leverage is available.
Overall, we expect the availability of third-party leverage and the terms of the leverage to
drive prices, especially for higher dollar-priced bonds.
Figure 19: Third-party leverage and yields for 2007 AAAs

Funding Unleveraged yield Repo leveraged yield
Sector Haircut cost Base Stress Base Stress
Prime Fix 2007 SSNR 15% L+150 7.7% 6.8% 29% 20%
Prime Hyb 2007 SSNR 20% L+150 8.2% 5.7% 27% 12%
Alt-A Fix 2007 SSNR 25% L+150 9.5% 7.8% 26% 18%
Alt-A Hyb 2007 SSNR 35% L+150 11.8% 8.3% 27% 16%
OA 2007 SSNR 40% L+150 10.5% 4.9% 20% 5%
st
Subprime 2007 1 Seq 20% L+150 5.6% -5.8% 11.4% -42.9%
Subprime 2007 LCF Seq 45% L+150 9.3% 1.1% 15.4% -4.0%
Source: Barclays Capital. Note: For leveraged yields, we assume that we roll financing every month and that forward
unlevered yields remain constant. Base scenario assumes that home prices fall 10% from here nationally. Stress
scenario assumes that home prices fall 30% from here nationally. Both scenarios assume an expected level of
modifications suitable for each housing scenario.
21 December 2009 52
PPIPs will play an important supporting role

PPIPs will provide a good The PPIP program announced last year will likely remain one of the major sources of
support level to prices support for both demand and prices in the non-agency sector in 2010. When the program
was first announced, it was expected to be one of the bigger sources of demand, since it
was the only program that had the benefit of leverage. With the availability of inexpensive
third-party repo leverage, prices in some sectors have rallied beyond the level at which it
makes sense for the PPIPs to get involved. However, the program will likely play an
important role in supporting prices in 2010.
Figure 20: PPIP managers’ status and targeted yields

Targeted
Manager Status Likely product mix
yield
Invesco Raised 20% 30% CMBS, 20/30/20% prime/alt-A/subprime

TCW Frozen/Mgmt Change
AllianceBernstein Raised 18% 40% CMBS, 45% alt- A, rest prime/subprime
Blackrock Raised
Wellington Raised 13-17% 35% CMBS, 25/25/15% prime/alt-A/subprime
Angelo Gordon & GE CAP RE Raised
Western Asset Management Raised
Marathon Raised 18% 15% CMBS, 30/20/35% prime/alt-A/subprime
OakTree Likely to delay funding All CMBS
Source: US Treasury, Connecticut Retirement Plans and Trust Funds
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
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.
Possible return of new issue deal creation in 2010

As a result of the strong rally in As a result of strong secondary market technicals and yield compression, we expect the new
secondary market prices… issue market to re-emerge in 2010 with relatively low volumes. Apart from a smattering of
deals in early 2008 and a couple of seasoned collateral deals in 2009, the new issue market
has been frozen since 2007. While there are still significant challenges to new deal creation,
with pending legislation and FDIC regulation potentially adding more, we are closer to new
deal creation from a pure deal economics standpoint today than we have been in the past
two years.
Closer to deal execution than in the past two years…

…we are closer to One reason is that risk-adjusted yields have dropped, especially on safer seasoned collateral,
deal execution economics’ to 4-6%. The emergence of the single-bond re-remic market in 2009 also showed that
making sense… “true” AAA bonds could still trade in the 6% yield context. While there will clearly be no risk
appetite for loans such as those originated in 2005-2007, we are likely to start seeing
enough demand for cleaner collateral - say, 60 LTV/CLTV, owner-occupied, full-doc, super
prime borrowers with low DTIs. These loans are currently being originated but mostly go
toward the banks’ lending portfolios. Just as TALF succeeded in bringing about new deal
creation in the ABS market by reducing the required yields on the top piece, we are likely to
see a similar effect from the availability of third-party repo funding on the non-agency
market. First, we will likely see a compression of unleveraged yields in the secondary
market. This should bring unleveraged yields down to levels at which new issue deal
creation would become probable. Furthermore, the leverage available on these new-issue
deals should make execution possible for very clean collateral deals. At present, agency
origination is a lot more attractive, with rates kept artificially low by the Fed. As the Fed exits
the agency MBS market, agency mortgage rates will rise and the compression in agency-
non-agency spreads will likely make new non-agency origination easier.
…but many legislative and regulatory obstacles persist

…but still have to surmount There are clearly plenty of obstacles to this process. The first is obviously getting yields low
hefty legal and regulatory issues enough to make deal economics work and make securitization an efficient way to fund
that are likely to crop up in 2010 these loans. The second obstacle is the FAS 166/167 accounting rules, which make it
difficult for banks to take the securitized assets off their balance sheets. While this does not
hamper individual securitizations, it makes it difficult for banks to churn out loans/deals
without affecting their leverage ratios. Other obstacles from the regulatory and legislative
changes in the pipeline could put further curbs on the securitization market. For example,
the FDIC and potentially Congress are likely to pass regulations/laws to force originators to
hold some proportion of loans (say, 5%) on their balance sheets. This, in conjunction with
the consolidation rules under FAS167, would make it extremely difficult for banks to sell
loans into a securitization from a capital relief or loss provision point of view. This is likely to
increase the costs of securitization as a source of funding and might make portfolio loans
that much more attractive for banks.
21 December 2009 54
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.
Figure 21: Cross-asset class unleveraged risk-adjusted yields

Sector Subsector Unleveraged yields
High grade 4.25-4.75%
Corporates
High yield 5.5-6%
Agency MBS Agency inverse IOs 11-14%
CMBS SD AAAs 5.7-8.5%
CMBS
CMBS AJs 5-12%
Consumer ABS CC 3y 1.8-2.3%

Auto 3y 2.0-2.2%
Jumbo SSNRs 8-9%
Alt-A SSNRs 9-12%
Non-agency MBS Option ARM SSNRs 8-10%
Subprime cash AAA 7-8%
Re-remic SSNR 5-7%
21 December 2009 55
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%
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
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
Jumbo FIX MEZZ 19.0 -16.8% 4.1% 17.8%

ARM MEZZ 18.0 -23.6% 3.6% 26.6%
Alt-A FIX MEZZ 18.0 17.4% 29.9% 38.8%
ARM MEZZ 11.5 3.7% 36.8% 59.2%
Negam NEG MEZZ 21.0 -83.5% -73.1% -45.8%
Note: 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.
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
1) Based on servicer advances

Subprime front cash flows will be When a delinquent loan is modified under HAMP, the servicer is expected to recapitalize the
hurt when servicers recoup P&I advances made and add this to the balance of the modified loans. From our discussions
advances on modified loans with servicers, we believe that they will reimburse advances from the top of the waterfall
once the loan is recapitalized and brought current through modification.
21 December 2009 57
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.
Long ABX 06-2 AAAs, Short ABX 06-2 PAAA

This trade can also be put on using the ABX. We already believe that ABX 06-2 PAAA versus
AAAs spread is too wide and needs to converge. This is because we believe that even with
modifications, we will start see crossover soon enough and the current 30 point spread
between the PAAAs and AAAs is unjustifiable. While some investors fear that any
widespread mods will delay crossover, we find that a combination of lower excess spread
due to mods and the advances recoupment by servicers once loans get modified are
enough to cause early enough crossover to justify a tighter price spread between ABX 06-2
PAAAs and AAAs. We believe that the spread should compress to the region of $23 and
recommend being long 1 unit of ABX 06-2 AAA versus 1 unit of ABX 06-2 PAAAs.
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%
40% Servicer modifies but

does not recoup P&I
20% FCF bond
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
2) Based on modification rates

Higher modification rates can
bring down deal CDRs As Figure 26 shows, HAMP modification rates can fluctuate wildly between servicers. The
December HAMP report also suggests that some servicers are modifying only portfolio
loans, as opposed to loans that are in securitized pools.
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%
15 Litton 19.8% 92% 2% 6%
High modification rate Avelo 16.0% 74% 2% 24%

10
Wells Fargo 13.2% 100% 0% 0%
5 Low modification rate
Countrywide 12.2% 100% 0% 0%
0 SPS 9.6% 49% 36% 14%
1 6 11 16 21 26 31 36 41 46 51 56 61 66 71
Month forward
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
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.
Figure 29: Prices at 10% yield across tiering scenarios
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
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
CMBS
Mind the gap(s)

Our macro outlook with respect to commercial real estate involves careful
Aaron Bryson
+1 212 412 3761
consideration of lag effects. We expect demand to begin to rebound across most
aaron.bryson@barcap.com
property types by H2 10, but property level income will continue to fall.
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.”
Quick recap of 2009

Before we dive into our outlook for 2010, we provide a brief recap of our views in 2009.
2009 held tremendous stress for commercial real estate and CMBS. However, despite the
headlines, the CMBS component of the US Aggregate Index posted 2,710bp of excess
returns, besting all major fixed income sectors with high volatility (Figure 1). This
apparent contradiction simply highlights the importance of what is discounted into
market pricing. Heading into 2009, most markets priced in a meaningful probability of a
21 December 2009 61
Figure 1: YTD excess returns versus Treasuries, volatility Figure 2: Estimated loss-adjusted yields
Bp 12% Loss/ 11%

6,000 10% extension 9% 9%
YTD 2009 XS returns vs Tsy (bp) 5,233 8%
adjusted 7%
8% 6%
5,000 Annualized volatility of monthly returns (bp) yields
6% 5%
4,000 4% 2% 2%
2,710 2%
3,000 2,357
2,072 0%
US HY
Credit card 5y AAA
Auto 3y AAA
Super sr jumbo AAA

CMBS LCF SD AAA
Subpime cash AAA
Super sr negam AAA
Super sr Alt-A AAA

US Credit
2,000 1,372 1,364
748
1,000 520 572
147 273 225
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.
Summary outlook for 2010

We still see value at the top of the capital structure. As shown in Figure 2, we see attractive
loss-adjusted yields in senior CMBS and residential credit versus other fixed income sectors.
We recommend investors capture this “structured finance risk premium,” that is the
additional yield offered versus less complex/distressed sectors such as investment grade
and high yield corporates. We forecast excess returns of 600bp versus Treasuries for the
CMBS component of the US Aggregate index, the bulk of which is from excess carry, but we
expect high volatility to persist.
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
Macro outlook: Last in, last out

Improving jobs data should help demand for space
Expect demand to turn positive In recent months, we have seen encouraging signs in the labor market, the single-best
by the end of the year across macro indicator for the demand for space across property types. Our economics strategy
most property types team forecasts sustained positive job growth starting in Q1 10, with more than 2.3mn jobs
added for the year. This corresponds to an unemployment rate of 9.1% at the end of 2010,
still a high figure. This will bring an end to the freefall in demand, or net absorption, across
most property types that has occurred since H2 08, with demand turning positive by the
end of the year (Figure 3). On the supply side, new completions will stay extremely low as
legacy projects are completed and loans for new projects are nonexistent.
But rents/income to respond with a lag

Especially in office, retail Despite a better jobs outlook, there will be a considerable lag between a rebound in the
demand for space and an actual improvement in rents and eventually income for most
property types. For example, in office sector during the prior two recessions, it took 2-3 years
after quarterly office employment turned positive before national office rents did (Figure 4).6
Exiting this recession, it should take several quarters of positive net absorption before
vacancies decline to a level at which landlords can have some degree of pricing power. Finally,
there is an additional lag between spot rents and property level income, given longer lease
terms in the office and retail sectors. We do not expect national office net operating income
(NOI) to turn positive until late 2013/early 2014. The one exception to the lags is hotel, the
sector with the shortest lease term, which has had the most immediate effect on cash flows
and for which we expect the quickest rebound.
And potential for secular shifts

Internet retailing gains share A unique aspect of this recession is potential secular shifts affecting many property types. On
retail, our macro forecasts call for a muted rebound in consumer spending, with private
consumption at only 2.1% in 2010, a historically tepid recovery. Another concern is the
Figure 3: Jobs outlook improving, should help demand for space Figure 4: But impact will be felt with considerable lags
Sq ft YOY % Chg Jobs (000s) Rent

300 (millions) forecast in NFP 4% 800 growth 4%
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
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
Multifamily stands alone

Demographic trends support The property type for which we can see the clearest rebound in demand is multifamily.
multifamily demand According to the Joint Center for Housing Studies at Harvard University,10 household
formation is expected to average 1.27-1.49mn per year in 2010-15, owing to the echo
boomers coming of age. We expect a greater share of these newly formed households to
rent compared with the recent past. This is supported by the recent decline in
homeownership rates from a peak of 69.2% in June 2004 to 67.6% as of September 2009;
since 1980, the average rate has been 65.8%. Another key difference for multifamily is the
availability of attractive financing, because of the presence of the GSEs. Of course, in ex-
housing bubble markets, this will be tempered by an excess of shadow single family and/or
condo supply.
Property values close to bottoming

Another theme for 2010 is that we expect CRE prices to bottom, with an increase in
We expect average peak to transaction activity as sidelined cash returns to the market to buy property. Our preferred
trough declines of 50% before a measure of property values is the Moody’s CPPI index, based on a repeat sales regression,
gradual rebound where values are already down 43% from the peak in October 2007. The decline is
widespread across property types (Figure 5). We find some interesting facts in the
performance of distressed versus non-distressed properties. Real Capital Analytics (RCA)
tracks troubled commercial real estate assets, including any property/loan with foreclosure,
bankruptcy, or restructured/modified status. Based on estimates from RCA and Professor
David Geltner from the MIT Center for Real Estate, “healthy” properties are down 38% from
the peak, versus 55% for “distressed” properties, with distressed accounting for 30% of
recent repeat sales data for the index (Figure 6). The price decline for the distressed
component has stabilized in recent months. As the percentage of distressed versus non-
distressed sales rise, we expect the index to continue to decline with a peak-to-trough fall of
about 50% by the middle of the year, before gradually rebounding in H2 10.
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
Banks and systemic risk

An outlook for CMBS in 2010 is not complete without a discussion of the banking sector.
Banks are easily the largest lenders against commercial property, comprising nearly 50% of
commercial mortgage debt outstanding. In Q3 09, banks/savings institutions had a $20bn
net contraction in CRE lending activity, the largest quarterly decline on record.
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.
Policy side: Life after TALF?

Another key theme in 2010 is likely to be the unwinding of policy support. As of now, the
legacy TALF program matures in March 2009 and the new issue TALF program in June
2009. The existence of legacy TALF provided a crucial backstop for the sector in 2009,
despite relatively limited participation. The general consensus is that both new issue and
legacy TALF will in fact expire at these dates. However, we believe there is a material
probability that the programs will be extended in 2010, particularly the new issue program.
In a December 10, 2009, letter to the Congressional Oversight Panel, Treasury Secretary
Geithner wrote, “We may increase our commitment to the … TALF, which is improving
securitization markets that facilitate consumer and small business loans, as well as
commercial mortgage loans. We expect that increasing our commitment to TALF would not
result in additional cost to taxpayers.” We agree and see little downside from extending the
program, especially because we have not seen any conduit deals come to market yet.
Figure 5: CRE values by property type Figure 6: Distressed versus non-distressed CRE prices
Index value 0%
200
-10%
175 -20%
150 -30%
125 -40% -38%

-43%
100 -50%
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
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
%, fixed 30+ day

12 rate delinquency
12%
universe rate 9.8%
10 30+ day deliquency rate = 6.5% 9.5%
10%
Specially serviced, current = 3.0% 8.0%
8
Total = 9.0% 8% 6.5%
6 6%
3.6% 3.8%
4 4%
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
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.
Figure 11: Average cumulative default chart, fixed rate CMBS
% of deal Fixed rate CMBS, cumulative default curve

2005+ vintage, cumulative default curve
60 2007 vintage, cumulative default curve
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.
21 December 2009 67
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
Figure 12: Estimate of upcoming CRE maturities
$ 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
Figure 13: 2009 CMBS maturing loans struggled Figure 14: Debt yield distribution of maturing loans
100% Over YTD 2009 period thru Nov., 56% of $ bn

maturing CMBS loans, net of defeasance, 160 MR NOI debt yield >=12%
90%
have paid off 140 MR NOI debt yield <12%
80%
20
70% 120 32
60% 51% 53% 54% 53% 100
45% 47%
50% 39% 38% 80 30
39% 37%
40% 33%
60 115
30% 18 102
40 21
20% 19 20 67
12
10% 20 39 30
20 26 21 9
0% 0 8
Jan Mar May Jul Sept Nov 2010 2011 2012 2013 2014 2015 2016 2017 >2017
3 mo paidoff Cumulative Paidoff Year of maturity
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.
Can modifications help?

In response to the income and funding gaps, we expect a wave of modifications in 2010.
We broadly classify potential modifications into two groups:
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
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.
Group B modifications promise to be much more complicated and scrutinized. In general,

we are sceptical about the efficacy of them. For many of these loans, the borrowers
extracted considerable equity at the peak of the market, and we suspect they will be
unwilling to commit significant new equity to the investment unless they get exceptional
rate reductions and/or principal write-downs from special servicers. In some of the more
positive cases, the mezzanine note investors will step into the first mortgage and attempt to
recoup some of their investment.
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
What about CMBS 2.0?

The restart of the CMBS market provides an opportunity to address many of these
concerns. To date, three deals have priced. We refer to these as CMBS 2.0, given their lower
leverage and more straightforward structure and underwriting. In addition, we see a variety
of different structural tweaks across these deals.
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.
Although encouraged by recent developments, we do not expect a surge in origination in

2010. We see multiple impediments, including uncertainty about risk retention legislation,
the effect of FAS 166/167 on the special servicer model, and concerns about warehousing
risk from loan origination to deal execution.
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
Base Base+ 2std dev Stress Stress+2std dev
21 December 2009 71
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.
Figure 17: Cumulative loss timing curve, fixed rate CMBS
% of deal Fixed rate CMBS, cumulative loss curve

2005+ vintage, cumulative loss curve
18 2007 vintage, cumulative losscurve
16
14
12
10
8
6
4
2
0
Note: As a % of cutoff balance. Source: Barclays Capital
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.
Technical pressures likely to rise

A potential source of technical selling pressure in 2009 has been rating agencies. To date,
downgrade activity in CMBS has been led by one agency. S&P has downgraded 46% of
21 December 2009 72
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.
In terms of the extent of downgrades, no agency has downgraded a duper to below

investment grade; the lowest rated duper is BBB- for GSMS 2007-GG10 by S&P. We believe
select dupers could be downgraded to below IG in 2010 by S&P, but would not expect other
rating agencies to be as harsh. For AMs and AJs, we suspect that Moody’s and Fitch will
follow suit, with several bonds downgraded to below investment grade in 2010.
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.
Rising rates could pose problem

Another key risk in 2010 is rising Treasury rates. Our rates strategy team forecasts the 10y
to reach 4.2% by midyear and 4.5% by year-end, versus 3.5% currently. This is above
market implied forward rates, and could negatively affect CMBS in multiple dimensions. The
first is on pricing, as rising rates would put upward pressure on cap rates and commercial
Figure 18: Loss-adjusted yields, senior AAAs Figure 19: Loss-adjusted yields, AMs & AJs
Yields (%) Yields (%)

10 9.2 30 27.1
8.5
9 7.9
8 7.1 6.9 6.8 25
6.6 19.1
7 6.0 5.9 6.0 5.7 5.6 5.8 5.7 5.6 20
6 14.5 13.5 14.4
15 12.7 12.9
5 10.3 10.6 11.5
8.8 10.1
4 8.68.2
10 7.0
3
2 5 1.9 1.0
1
0
0
Old Super Last Cash Last Cash Last Cash -5 -2.7
School2005 Sr2005 Flow2005 Flow2006 Flow2007 AM2005 AM2006 AM2007 AJ2005 AJ2006 AJ2007
0 CDR Base Stress 0 CDR Base Stress
12
U.S. Life Insurers’ Commercial Mortgage Exposure & Losses are Manageable, Moody’s Global Insurance Special
Comment, December 2009.
21 December 2009 73
Figure 20: 2005+ vintage LCF dupers, % D/G by count Figure 21: 2005+ vintage AMs/AJs, % D/G by count
120% 120% Still AAA

Negative Downgraded, IG
100% 2% Downgraded, Non-IG
Watch 100%
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
S&P Moodys Fitch AM AJ
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.
CMBX relative value trades

See mispricing across CMBX Finally, a growing focus for us in 2010 will be to exploit mispricing across the CMBS capital
series/capital structure structure. All of our trades are grounded in our fundamental, loan-level analysis. We already
mentioned our favorite outright short in CMBX, A.3. We also see less beta-oriented relative
value trades with considerable upside. Our favorite such trade is to buy AJ.2, sell AJ.5, where
we believe the difference in collateral quality between Series 2 and 5 is not adequately
reflected in current pricing. We see 7-8 points of potential upside (Figure 22). We also
recommend buying A.4 versus selling A.3, which is a positive net upfront, positive carry
trade with 3-4 points of potential upside (Figure 23).
Figure 22: Buy AJ.2, sell AJ.5 Figure 23: Buy A.4, sell A.3
70 AJ2-AJ5 px 40 A4 - A.3 px (rhs) 8

AJ.2 A.4
60 AJ.5 A.3
35 6
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
21 December 2009 74
CONSUMER ABS
One year later, the world is a better place

Joseph Astorina At least in consumer ABS, that is. A year ago, this market was on the precipice:
+1 212 412 5435 future collateral performance looked horrible, spreads were at all-time wides, and
joseph.astorina@barcap.com the financial system was teetering on the brink. Heading into 2010, performance is
stabilizing, spreads are tightening, and the US economy and financial system are on
the mend.
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.
Themes for 2010

While we do not expect the heady returns earned in consumer ABS during 2009, we believe
2010 will still provide ample opportunity for savvy investors. Among the major themes we
believe will dominate are regulatory and legislative risk, adjusting to life after TALF,
stabilization in collateral performance, and a continuation of the rally in spreads.
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
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
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:
Setting or limiting interchange fees
Requiring disclosure of interchange fees to consumers
Restricting card networks from imposing rules on merchants (eg, preventing merchants
from imposing a surcharge for using credit cards)
Allowing merchants and issuers to negotiate rates directly.
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
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.
Life after TALF

The consumer ABS market The Federal Reserve’s Term Asset-backed Lending Facility (TALF) was a leading example of
appears ready to wean itself off a 2009 government intervention that worked as intended. As Figures 1 and 2 clearly
TALF demonstrate, TALF was instrumental in re-starting the consumer ABS new issue machine in
2009 after a moribund close to 2008. TALF-eligible deals comprised the lion’s share of new
issuance in 2009. Nevertheless, over the course of the year, non-TALF transactions also
trickled into the market. New issue volume in the major consumer ABS sectors (auto, credit
card, equipment, and student loan) totalled $128bn, compared with $125bn in 2008. Of
that, roughly 70% was TALF eligible. However, despite being TALF-eligible, many of the
transactions in the waning months of 2009 were technically not “TALF-able” (ie, pricing
spreads were too tight relative to TALF financing rates). Approximately $50bn of TALF loans
were requested in 2009, representing about 55% of TALF eligible consumer ABS issuance.
However, these numbers include an undisclosed amount of TALF refinancings, rather than
new borrowings, and likely overstate the utilization of the program. Regardless, the rebound
in new issue volume (albeit well below historical levels) and the fact that most issuers no
longer need TALF to bring a transaction, combined with the spread tightening over the
course of the year, make TALF an exceptional success, in our view.
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.
Figure 1: TALF-eligible ABS issuance Figure 2: Non-TALF-eligible ABS issuance
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
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 $15-20bn for student loans.
Figure 3: Consumer ABS issuance in 2009, by asset type ($mn)
Card
42,671
33.3%
Auto
58,198
45.4%
Student Loan
Equipment 18,062
9,216 14.1%
7.2%
Source: IFR Markets
Stabilization in collateral performance

Although collateral credit deteriorated throughout 2009, consumer ABS performance held up
well. We believe 2010 will bring stabilization in collateral performance, albeit at still high levels,
as the economy recovers and consumers rebuild their balance sheets. We maintain our peak
aggregate credit card charge-off forecast of 11.0-11.5% in Q1 10, coinciding with our
economists’ call on peak quarterly unemployment (10.1% in Q1 10). Subsequently, we see
charge-offs moderating slightly, settling into 9-10% for the balance of 2010. Retail auto loan
collateral performance has already shown signs of stabilizing; seasonally adjusted
delinquencies and annualized net losses have stopped increasing. Nevertheless, we do not
believe auto loan performance is likely to return to levels prior to the mortgage crisis (mid-
2007) any time soon. Rather, delinquencies and annualized net losses on retail auto loans are
likely to stay rangebound in 2010 (0.5-0.8% and 3.5-5.0% for prime and non-prime
delinquencies, and 1.6-1.9% and 8.0-11.5% for prime and non-prime annualized net losses).
21 December 2009 79
Spreads and trade recommendations

Throughout 2009, consumer ABS spreads across the capital structure tightened
significantly as investors looked into the abyss, regained comfort with the credit story, and
were enticed back to the market with historically attractive yields. Despite the significant
rally (spreads are currently well inside levels during the depths of the credit crisis, post-
Lehman bankruptcy), spreads are still well wide of pre-mortgage crisis (ie, prior to July
2007) levels. We believe they have the potential to continue tightening toward historical
levels, albeit likely settling somewhat higher. We do not expect them to return to pre-July
2007 levels anytime soon, nor do we expect them to remain this wide. Our view is driven by
the expected tight supply of consumer ABS in 2010, which, when combined with the large
amount of investor cash on the sidelines, will likely result in continued spread tightening on
consumer ABS.
Figure 4: Consumer ABS spreads: Current versus 2y average pre-Lehman bankruptcy +/- one standard deviation
850 Average, 2y pre-Lehman Bk Current Level
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 (A) Subordinate (BBB) AAA

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
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.
Auto ABS outlook

Auto-related ABS issuance totaled $58.2bn in 2009, a 59% increase over 2008’s $36.5bn
and 15% below the $68.8bn in 2007 (Figure 5). Prime and non-prime retail auto transaction
volume was up 15% and 10%, respectively. However, thanks largely to TALF, the growth
engines in consumer ABS were the less common (historically) asset classes, such as retail
lease ($7.7bn in 2009 versus $550mn in 2008), dealer floorplan ($2.4bn versus $0), rental
car ($2.1bn versus $0) and fleet lease ($2.6bn versus $0). In the absence of government
intervention, the primary market for auto ABS would likely have stumbled into 2009 as
dismally as it ended 2008. In fact, prior to the inception of TALF, only two retail auto ABS
transactions from high quality issuers priced. However, issuance picked up quickly once
TALF got rolling.
Figure 5: Auto ABS issuance rebounded in 2009, but remains below historical levels
Prime retail loan Non-prime retail loan Retail lease

120 Dealer floorplan Motorcycle loan Rental car
Fleet lease
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
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
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
21 December 2009 83
Credit card ABS outlook

Without a doubt, regulatory and legislative risks are the largest issues overhanging the
credit card ABS market in 2010. Both issues served to slow issuance in this sector to a
trickle in Q4 09 and will likely continue to act as a drag on volume in 2010. Credit card ABS
issuance totaled $42.7bn, down 26% from $57.5bn in 2008 and a record $90.9bn in 2007
(Figure 11). As with auto ABS, absent government intervention, issuance of credit card ABS
in 2009 would have been even lower. Notably, retail card ABS had a strong showing,
accounting for 37% of the total dollar amount securitized.
Figure 11: Credit card ABS issuance declined on FAS 166/167, FDIC uncertainty
100 Bank card Retail card

90
80
70
60
50
40
30
20
10
0
2001 2002 2003 2004 2005 2006 2007 2008 2009
Source: IFR Markets
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 $35-40bn of new credit card ABS, the bulk
of which will likely be from bank card issuers, with the remaining maturities financed some
other way. In addition, a whopping 90% of maturities are senior class A notes, so there will
likely be less pressure to issue subordinate classes (publicly or privately) to maintain
required enhancement levels.
21 December 2009 84
Figure 12: Credit card ABS roll-off in 2010 ($bn)
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
Change in Charge-offs Change in Charge-offs

6 5.0
2
5 2 4.5 R = 0.7996
R = 0.7017
4 4.0
3 3.5
3.0
2
2.5
1
2.0
0 1.5
-1 1.0
-2 0.5
-3 0.0
-1 0 1 2 3 4 5 -1 0 1 2 3 4 5
Change in Unemployment Rate Change in Unemployment Rate
Source: BLS, Moody’s Investors Service, Barclays Capital Source: BLS, Moody’s Investors Service, Barclays Capital
21 December 2009 85
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
Student loan ABS outlook

Despite Congress’ and the Obama Administration’s best efforts to eliminate the FFELP
program, it has managed to survive, for now; the issue of closing it down and having the
government provide education loans directly to students will come up again in 2010.
Indeed, Secretary of Education Arne Duncan recently penned an op/ed piece arguing
against banks’ role in student lending.
21 December 2009 86
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
70 Sallie Mae Nelnet Brazos Student Loan Corp.

First Marblehead College Loan Corp. Collegiate Funding Access Group Inc.
Goal Education Funding Keycorp
60
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
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
21 December 2009 88
SPECIAL TOPIC
Housing risks and prospects

Glenn Boyd Home prices showed remarkable strength in Q2-Q3 09, buoyed by declining distressed
+1 (212) 412 5449 inventories, as well as low mortgage rates, the first-time homebuyer tax incentive, and
glenn.boyd@barcap.com summer seasonals. In 2010, all of these effects will likely change. Mortgage rates could rise
as much as 100bp, based on our rates and agency basis calls; the tax incentive will expand,
Robert Tayon then expire; seasonal distressed distortions and direct distressed inventory should evolve.
+1 212 412 2512 How quickly distressed inventories rise depends on near-term servicer foreclosure practices,
Robert.Tayon@barcap.com which can be difficult to predict. Our REO/HPA model suggests a home price decline of 8%
from Q3, bottoming in early Q2 10. The primary drivers for this projected decline are
abnormally strong seasonals, along with expectations of rising REO inventory.
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
HPA in winter 2010, if this seasonal pattern holds. That is 20% weaker HPA than has
recently been displayed.
Of course, we do not expect underlying (distressed-excluded) HPA to be unchanged this

winter. However, the First American Core Logic’s Loan Performance HPI data provide
additional evidence that abnormal seasonals will depress home prices in the coming
months. This is the major reason we expect home prices to fall in Q4 and Q1.
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
Figure 1: Abnormal seasonals across states
ALL - DX (3m avg) Individual states Aggregate

25
20
15
10
5
0
-5
-10
-15
-20
Jan-05 Jan-06 Jan-07 Jan-08 Jan-09 Jan-10
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
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.
Unemployment, rates, and taxes

…but the HPA/REO relationship While distressed inventories should dominate HPA in bust regions, they cannot explain
holds up mainly in housing bust dynamics in non-bust regions. Even bust regions will gradually transition to non-bust
states, not in non-bust areas dynamics over time, and more traditional variables must be invoked to broaden the
applicability of the model. In addition, regulatory changes such as the first-time homebuyer
tax incentive can be translated into an effective rate reduction, highlighting the importance
of a rate component. Thus, our updated regional HPA model incorporates unemployment,
rates, and the tax incentive (modeled as a rate change).
Rates are relative

As bust regions transition to a In an equilibrium view, in which constant rates imply constant prices, only changes in rates
non-bust environment, rates, the should affect HPA. However, experience shows that borrowers react to low rates over a
jobless rate, and tax incentives period of several months, raising the issue of the time scale for rate changes. Intuitively, a
should start to play a bigger role purchase decision shares similarities with a refinance decision – new record lows spur
marked activity, whereas rates that are high from a historical perspective spur little more
than yawns. This motivates us to use rate attractiveness – the difference between 3-year
average and current mortgage rates – as the independent rate variable. Rate attractiveness
or similar media variables are regularly used to model refinancing waves and have a
demonstrably strong correlation with prepayments over decades.
Tax incentives can be converted into rates

Our model converts the tax To incorporate the tax incentive, we translate savings into an effective rate by amortizing
incentive to a rate incentive of $8,000 on a $150,000 loan over four years. Estimating roughly one-third of potential
about 44bp homebuyers as qualifying first-timers gives an effective rate reduction of 44bp. The
expanded incentive includes higher-income limits. Income distributions and
homeownership rates suggest about an additional 10% of qualifiers, but we assume
$200,000 average loans, resulting in a rate-equivalent change of just over 3bp. The larger
Figure 2: CA HPA and distressed inventories Figure 3: CA HPA, unemployment, and rate attractiveness
HPA (nsaar) REO as %stock HPA (nsaar) Adj Rate Attractiveness

50 2.0 30 200
40 20
150
30 1.6 10
20 0 100
10 1.2 -10
0 50
-20
-10 0.8
-30 0
-20
-40
-30 0.4 -50
-40 -50
-50 0.0 -60 -100
HPA -250*REO chg REO HPA -20 *Une chg RA

Source: First American Core Logic’s Loan Performance HPI, Barclays Capital Source: First American Core Logic’s Loan Performance HPI, BLS, Barclays Capital
21 December 2009 91
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
HPA = a + r1*REO + r2*dRdt + u*dUdt + ra*RA + sv1 +SV + sv2*SV*REO,
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
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).
Scenarios and results

The pace of foreclosures slowed to a crawl in 2009, but increased marginally in October. As
a result, REO inventories rose for the first time all year. Historically, foreclosure-to-REO
(F2R) roll rates averaged 8-10%, but are now below 4%. Our base case corresponds to
slight increases in F2R roll rates, to roughly 5%. Figure 4 shows the base-case REO levels for
California. We also consider an optimistic scenario in which F2R remains near current levels
throughout H1 10 and demand (in particular, REO liquidation rates) grows 30% by next
spring. For our stress scenario, we consider what happens if servicers start moving
delinquent loans into REO more aggressively, with F2R increasing to roughly 6%.
We do not have a state-level unemployment model, so we make the very simplistic

assumption that unemployment rates decline 0.1 points each month after Q1 10, scaled by
the ratio of [current state-level unemployment]/[current US unemployment], for the next
four years. While this clearly leaves much to be desired, our regressions are not strongly
dependent on unemployment unless it jumps 0.5 or more points (as occurred last winter).
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
Figure 4: CA distressed inventory scenarios Figure 5: CA projected home prices
REO (%stock) HPI

1.6 105
100
1.4
95
1.2
90
1.0 85
stress 80 stress
0.8 base base
75
optimistic optimistic
0.6 70
US and key states

Among other states, Nevada is We outline projected remaining depreciation (from Q3) for key states in Figure 6. The model
very vulnerable with a further expects extraordinarily weak performance in Nevada, with home prices declining another
decline of 20-30% expected 30% in the base case. The reason is the high density of foreclosures: at 2.2%, Nevada has
roughly twice as high an REO level as California. In this case, the model may be too punitive.
We have not included long-term equilibrium sub-models, largely because we are focused on
1-3 year horizons and partly because price-to-income ratios are roughly in line with long-
term historicals on a nationwide basis. In Nevada, that deficiency ignores a potential pull to
equilibrium, which could result in a less severe downturn. We hope to incorporate
equilibrium aspects at a future point, but the issue reinforces our general view that models
are starting points and should be adjusted with subjective corrections when appropriate.
Nevertheless, the outlook for Nevada appears bleak, with a 20% or larger decline likely.
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
Figure 6: Projected remaining depreciation Figure 7: US aggregate HPI
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
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
SPECIAL TOPIC
Agency underwriting: When will the squeeze ease?

Nicholas Strand As home prices and the economy recover, mortgage credit should eventually loosen.
+1 (212) 412 2057 However, the timing, as well as the ultimate renormalized level, is uncertain.
nicholas.strand@barcap.com We suggest that risk aversion, loan repurchases, economic uncertainty, and
regulatory scrutiny will gradually give way to the need to build origination volumes
Glenn Boyd and increase profits.
+1 (212) 412 5449
glenn.boyd@barcap.com While it seems unlikely that credit will loosen over the next 3-6 months, certain
credit segments should see increased availability over the next 6-12 months.
In particular, in H2 10, there could be a meaningful extension of conventional credit

to currently underserved segments: the substantial population of borrowers with
low LTV but only mid-range FICOs (700-740).
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
780 40% 16%

760 35% 14%
740 12%
30%
720
10%
700 25%
8%
680 20%
6%
660
15%
640 4%
620 10% 2%
600 5% 0%
Jan-06 Oct-06 Jul-07 Apr-08 Jan-09 Oct-09 600 640 680 720 760 >800
FICO
FICO (FHLMC) FICO (FHA) Pct with LTV>80
2007 2009
Source: HUD, Freddie Mac, Barclays Capital Source: Freddie Mac, Barclays Capital
21 December 2009 96
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.
Reduce seller concessions: Currently, sellers are permitted to contribute up to 6% of

the purchase price of the house to cover closing costs or other expenses. The FHA
would like to reduce this to only 3%.
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.”
Justifications for tight credit

In examining credit standards across a range of large and mid-sized originators, three
common themes emerge. We believe the main reasons for sparse mortgage credit
availability are restrictive GSE automated underwriting; potential losses from repurchases;
and general risk aversion and uncertainty about housing and the broader economy.
GSEs have tightened credit GSE Underwriting

through increased risk based Over the past couple of years, the GSEs have sharply adjusted their underwriting and risk-
pricing, revised debt-to-income based pricing in response to the deteriorating housing market and rising delinquencies on
targets, and increased their guarantee business. While hardly exhaustive, we list a few relevant changes:
documentation requirements
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
Increased up front fees for high LTV/low FICO borrowers (LLPAs)
Reduced maximum debt-to-income requirements
Increased documentation requirements
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.
In terms of credit availability, changes to documentation and debt-to-income requirements

Revised debt-to-income limits
seem to be the main reasons on the GSE side for the dramatic reduction in the agency
and increased documentation
underwriting box. Several key changes announced by Freddie Mac in its Single-Family
requirements seem to be a
Seller/Servicer Guide Bulletin, November 200817 seem to be at least partially responsible for the
bigger deal
tightening in agency credit in late 2008/early 2009 (roughly a 30bp increase in FICO as seen in
new agency originations). Among the significant changes announced in this bulletin:
Decrease the maximum debt-to-income ratio to 45% for most mortgages
Elimination of stated income/stated asset mortgages
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
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
21 December 2009 99
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
2008 2009 BAC C JPM USB WFC
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.
21 December 2009 100

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.
Other risk-aversion aspects, such as bias by momentum-driven appraisers, should abate as

housing stabilizes. More generally, we believe that the scramble to contain losses and risks
has led to an over-correction in mortgage documentation and credit standards. This should
reverse over the next 6-12 months, as uncertainty in housing and economic conditions
shrinks, investor risk tolerance grows, bank earnings share the stage with losses, and the
government places increasing pressure to lend. Credit will not balloon overnight. Rather, the
next cleanest credit segments are likely to be the first beneficiaries, such as mid-FICO, low-
LTV borrowers.
Origination economics suggest a move to lower FICOs is inevitable

Originators have an economic The most visible reason for tight credit is the ballooning repurchases on legacy loans. We
incentive to start issuing more believe that quality controls on documentation, combined with continued prudence on
mid-tier FICO loans leverage (LTV and affordability products), should greatly reduce this risk on new origination.
But in the event of a second leg down, mid-FICO borrowers will clearly default at higher
rates than the high-FICO borrowers typical of current origination. Why, then, should lenders
ever move down in FICO? Every bank desires higher origination volumes, but moving down
in FICO is sensible only if the risk-adjusted return is adequate. A simple analysis suggests
that not only are mid-FICO loans economically sensible, they should be even more
profitable than high-FICO origination in a forward-looking environment.
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 seem poised to decline

Freddie Mac reports it put $2.7bn in mortgages back to lenders from Q1 to Q3 09. Assuming a
4.5% serious delinquency rate on its $1.6trn guarantee portfolio, this represents an annualized
repurchase rate of 3.8% of seriously delinquent loans. The expected repurchase cost on a new
loan can be described as the probability of becoming seriously delinquent times the likelihood
of being repurchased. Figure 10 shows that roughly 9% of prime-quality, mid-FICO loans
(700-750 FICO) were seriously delinquent by 36 WALA in 2009, compared with only 4% on
high-FICO (750-800 FICO) loans. Adding in liquidated loans, we find cumulative non-
performers of approximately 13% on mid-FICO, versus 6% on high-FICO. Thus, mid-FICO
non-performance was about 7 points higher than high-FICO in 2009. Assuming a 3.8%
repurchase rate and a loss of 60 cents on each repurchased loan, we estimate (7%) x (3.8%) x
(0.6) = 16bp marginally higher loss ($0.16 for every $100) for mid-FICO loans.
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
Source: LoanPerformance, Barclays Capital Source: LoanPerformance, Barclays Capital
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.
Convexity benefits overwhelm the costs of credit

The main benefit from originating mid-tier FICO loans comes from the superior prepayment
behavior of these borrowers, which leads the MSR to be valued at a higher multiple. Figure
11 shows refinancing response curves over the past six months by FICO for loans that are
12-24 WALA and less than 75 current LTV. During this period, deep-in-the-money speeds
for 760-790 FICO borrowers were nearly twice as fast as those borrowers with less than 700
FICO. What these data clearly show are that the increasing FICO hurdles placed on new
originations have diminished borrowers’ ability to refinance.
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
1M CPR Refinancing Multiple

50 1.4
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
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.
Figure 13: MSR valuations to benefit from lower FICO loans

% of Refi Model
5% MSR Proxy
100% 90% 80% 70% 60% 50%
Price 20-26+ 21-17 22-9+ 23-4+ 24-02+ 25-04

Multiple 4.17 4.31 4.46 4.63 4.82 5.03
Additional Multiple 0.00 0.14 0.29 0.46 0.65 0.86
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
Subprime speeds: Assessing credit curing

Glenn Boyd While a lack of refi options has temporarily depressed subprime speeds, we consider
+1 212 412 5449 the catalysts for a recovery.
glenn.boyd@barcap.com
Agency underwriting standards should decline from currently elevated levels in 6-18
months, restoring the ability of some subprime borrowers to cure. This base-case
Robert Tayon
expectation can boost yields 150-250bp.
+1 212 412 2512
Robert.Tayon@barcap.com We expect a return to $50-70bn of low LTV subprime issuance per year, perhaps as
soon as the next 18-30 months. Such a scenario represents considerable upside
(100-400bp in yield) from the base case.
Speeds are likely to accelerate modestly, even in conservative scenarios.

Considerable prepayment upside in a prolonged recovery environment confers
additional optionality, making subprime an attractive sector.
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
Figure 1: Subprime refis are now dominated by the GSEs
New Issuance CRR contribution

subprime
50
prime
45
Alt-A
40 FHA
35 FN/FH
30 Actual CRR
25
20
15
10
5
-
Jan-04 Jan-05 Jan-06 Jan-07 Jan-08 Jan-09
Source: LoanPerformance, FNMA, FHLMC, FHA, Barclays Capital

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.
Scenarios for credit expansion

High credit spreads at origination provide a strong refi incentive for subprime borrowers to
cure. Historically, credit curing has been in the form of a reduction in LTV from home price
appreciation and/or improved FICO scores from timely payments. Because typical subprime
borrowers have SATOs of 200bp or more, prepayments historically depended far more on
credit curing than on rates. Recent slow speeds reflect the elimination of credit to this
demographic and a drastic reduction even for subprime borrowers who cure enough to
bring FICOs to 680, 700, or even higher. While the credit environment is likely to take years
to equilibrate, we see three potential drivers of increased prepayment speed: relaxation of
agency FICO standards; low-LTV, low FICO issuance; and home price appreciation.
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.
Figure 2: Prepayment speeds by OTS status Figure 3: Prepayment speeds by LTV
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
AC CPD 30-59 LTV<80 80-90 90-100 100-110

Note: AC: always current; CPD: current, previously delinquent; 30-59: days Source: LoanPerformance, Barclays Capital
delinquent. Source: LoanPerformance, Barclays Capital

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.

Quantifying the upside

Even with expanded credit availability, not all borrowers will qualify. We focus on LTV and
delinquency status as the key determinants of prepay speeds. The composition of different
vintages varies widely in these factors; for example, due to price appreciation from 2004 to
2006, seasoned vintages typically have a greater proportion of low-LTV, always current
borrowers (Figure 4). These borrowers would be the most likely to prepay if credit
conditions improve. On the other hand, recent vintage bonds trade at a deeper discount, so
each prepayment is more valuable to the investor. Across a basket of penultimate and LCF
subprime bonds, we find that upside scenarios involving only credit relaxation affect recent
and seasoned vintages comparably, while those involving HPA improve recent vintages
more than seasoned.
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
Figure 4: Distribution of prepay candidates by vintage Figure 5: Simulated CRR scenario
40% 2004 12 2004

35% 2007
2007 10
30%
25% 8
20% 6
15%
4
10%
5% 2
0%
0
LTV<80 LTV 80-90 LTV<80 LTV 80-90
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.
Figure 6: CRR upside

Scenario 2004 vintage 2007 vintage
Current CRR 4 2
H207 conditions 11 5
H207 + new origination 17 11
H207 + orig. + 10% HPA 18.2 13
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.
Figure 7: Loss-adjusted yields for representative bonds

CBASS CBASS
HASC 2007 SVHE 2006 SVHE 2006 2004 CB6 2004 CB6
Scenario NC1 A3 EQ1 A3 EQ1 A4 AF3 AF4
Current CRR 13.2 15.3 16.1 9.2 10.1
H207 conditions 15.2 18 17.6 10.1 12.2
H207 + new
origination 18 20.7 20.2 10.9 13.7
H207 + orig.
+ 10% HPA 19.2 22.8 20.6 11.1 14.1
Note: Based on indicative prices as of December 17, 2009. Source: Barclays Capital
To estimate the price effect, we look at even-yield price changes from the yield in the
current scenario (Figure 8).

Figure 8: Even-yield prices from current CRR yield

HASC 2007 SVHE 2006 SVHE 2006 CBASS 2004 CBASS 2004
Scenario NC1 A3 EQ1 A3 EQ1 A4 CB6 AF3 CB6 AF4
Current CRR 34 44 36 67 67
H207
conditions 36.3 48.3 38.5 71.9 75.3
H207 + new
origination 39.4 52.1 42.3 75.4 78.8
H207 + orig. +
10% HPA 40.6 54.5 43 76.1 79.6
Note: Based on indicative prices as of December 17, 2009. Source: Barclays Capital
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.

US SECURITIZATION STRATEGY ANALYSTS

Ajay Rajadhyaksha
Head of US Fixed Income and
Securitised Products Strategy
+1 212 412 7669
ajay.rajadhyaksha@barcap.com
Joseph Astorina Sandeep Bordia Glenn Boyd Aaron Bryson

ABS Strategy Residential Credit Strategy ABS Strategy CMBS Strategy
+1 212 412 5435 +1 212 412 2099 +1 212 412 5449 +1 212 412 3761
joseph.astorina@barcap.com sandeep.bordia@barcap.com glenn.boyd@barcap.com aaron.bryson@barcap.com
Yan Cao Derek Chen Tee Yong Chew Sandipan Deb

Agency MBS Strategy Agency MBS Strategy CMBS Strategy Residential Credit Strategy
+1 212 412 5996 +1 212 412 2857 +1 212 412 2439 +1 212 412 2956
yan.cao@barcap.com derek.chen@barcap.com teeyong.chew@barcap.com sandipan.deb@barcap.com
Wei-Ang Lee, CFA Philip Ling Keerthi Raghavan Siddarth Ramkumar

ABS Strategy Fixed Income Strategy Residential Credit Strategy Agency MBS Strategy
+1 212 412 5356 +1 212 412 3202 +1 212 412 7947 +1 212 412 7581
weiang.lee@barcap.com philip.ling@barcap.com keerthi.raghavan@barcap.com siddarth.ramkumar@barcap.com
Matthew Seltzer Nicholas Strand Robert Tayon Jasraj P. Vaidya

Agency MBS Strategy Agency MBS Strategy ABS Strategy Residential Credit Strategy
+1 212 412 1537 +1 212 412 2057 +1 212 412 2512 +1 212 412 2265
matthew.seltzer@barcap.com nicholas.strand@barcap.com robert.tayon@barcap.com jasraj.vaidya@barcap.com
Saravanakumar Velayudham Elena Warshawsky

Agency MBS Strategy Residential Credit Strategy
+1 212 412 2099 +1 212 412 3661
kumar.velayudham@barcap.com elena.warshawsky@barcap.com

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.
21 December 2009 LAST PAGE

CMBS

Enviado por

Dados do documento

Descrição original:

Título original

Direitos autorais

Formatos disponíveis

Compartilhar este documento

Compartilhar ou incorporar documento

Opções de compartilhamento

Você considera este documento útil?

Este conteúdo é inapropriado?

Direitos autorais:

Formatos disponíveis

CMBS

Enviado por

Direitos autorais:

Formatos disponíveis

SECURITIZATION RESEARCH

U.S. SECURITIZED PRODUCTS

Opposing forces: Prepayments in 2010 21

Top trades for 2010 29

Agency underwriting: When will the squeeze ease? 96

Sub-prime speeds: Assessing credit curing 106

Back from the brink

Stepping back from the limelight

Rates markets and the Fed: A benign macro picture

Source: US Treasury, Flow of Funds, Barclays Capital Source: Barclays Capital

Homes (000s) Cum Delinquency

0-10 10-20 20-30

Figure 7: Bank Losses over the next few years ($bn)

1st lien mortgages 1741 11% 200 75 63 63

Problem areas: CMBS and tight residential mortgage credit

Sq ft (millions) YOY % Chg in NFP 770

Lower systemic risk – Making money through relative value in 2010

Things to watch out for, and risks to beware of

It is easier to comment on liquidity requirements. Regulators worldwide will force banks to

Mortgage basis outlook: Life after Fed

200 2006 2007 2008 2009 YTD

Review of 2009: Basis tightening and index outperformance

As a consequence of this supply-demand imbalance, the mortgage basis tightened

Tradable float has been significantly reduced

Implied funding for mortgages is at historical lows

Mortgages traded to short empirical durations in 2009

10 4.5s 5.0s 5.5s 6.0s

4 Empirical Hedged Carry 4.4 6.6 7.4 7.6

Roll Richness 1.7 2.1 2.5 1.8

Source: Barclays Capital Source: Barclays Capital

The Fed and agency MBS in 2010

Supply: Few changes from 2009

Net issuance ($bn) 450

800 Serious delinquency rate

Demand side: The usual suspects

20% 2.5 50,000

10% 1.5 30,000

-10% -0.5 (10,000)

Foreign demand: Risk reversal

Source: FDIC, Barclays Capital Source: FDIC, Barclays Capital

Long-term securities 6,262 7,162 9,136 9,463 8,792 8,212 9,528

Money managers: Entry at wider spreads

Supply % in Agency MBS

Combined $344.0 -23%

Source: Barclays Capital Source: Barclays Capital

Basis call: A stable basis at wider levels

Things to watch out for

Opposing forces: Prepayments in 2010

 For GNMA, continued tightening in underwriting by the FHA should reduce

A slow upward trend for voluntary prepays

50 Dec 2009 report

Ultimately, the return of refinance-ability depends on the recovery of HPA, normalization of

Buyouts to surge, causing significant tiering in speeds

CPR Involuntary prepayments (CPR) Modification CDR

Although FNMA is required to repurchase a loan once it becomes 24 months delinquent,

Figure 6: FHLMC/FNMA 1y prepayment projections

GNMA/FNMA speed differential to compress

 Increase the minimum FICO score

 Increase the down-payment requirement

 Increase the upfront and annual mortgage insurance premium (MIP)

 Increase lender accountability and indemnification for wraps and warranties

Turnover to slow to 4-5 CPR, GNMA to provide extension

For GNMA, continued tightening in underwriting by the FHA should reduce

Increase the minimum FICO score

Increase the down-payment requirement

Increase the upfront and annual mortgage insurance premium (MIP)

Increase lender accountability and indemnification for wraps and warranties

Coupon swaps hedged with Treasuries – Similar to a long IO position, up-in-coupon

Mortgage options – Mortgage options allow investors to implement conditional spread

Base case – Production model with no modifications or changes.

Muted voluntary speeds – We reduce voluntary prepayments to reflect the current

Forgiveness instead of forbearance: Debt forbearance can reduce payments for

Streamlining documentation requirements: Current HAMP guidelines require a