(Lehman Brothers) Mortgage Outlook For 2007 - Bracing For A Credit Downturn

Fixed Income Research U.S.
Securitized Products Research
December 12, 2006
Mortgage Outlook for 2007: Bracing for a Credit Downturn

OVERVIEW 2 As suggested by our bearish title, credit is set to dominate headlines in 2007. We expect mortgage collateral performance to deteriorate significantly next year and, accordingly, we have an overall bias to be up-in-credit. FLOW PRODUCTS
5
Srinivas Modukuri 212-526-8311 Prime MBS Vikas Shilpiekandula Prasanth Subramanian Jose John Eric Wang Olga Gorodetsky Sub-Prime MBS Akhil Mago Jasraj Vaidya Rahul Sabarwal ABS CDOs Michael Koss Dan Mingelgrin Prepayments Stefano Risa Phanwadee Khananusapkul
Demand technicals for mortgages should remain strong but we recommend a moderate underweight to fixed-rate mortgages versus a barbell of hybrids and CMBS. Our other recommendations for the year include an overweight to 15yrs and sourcing call protection through various ways including through GN/FN swaps, LLB 6s and new WALA 6.5s. PRIME MBS 30
We like owning hybrids due to a combination of relatively attractive valuations and a favorable supply/demand outlook. Own hybrids versus premium fixed rate MBS as well as swaps. We also favor selling liquidity through a core long position in the nonagency basis. In addition, we like call protection stories in structured MBS. In credit, we recommend moving up-in-borrower-quality and down-in-rating. NON PRIME MBS 45
Credit concerns regained focus in 2006 and are expected to intensify in 2007 resulting in significant negative headlines in the sector. While we advocate an overall defensive posture in 2007, we continue to expect tiering to increase in 2007 based on cross-vintage, capital structure, and credit views, which should provide significant opportunities for intra-sector relative value. ABS CDOS 51
Coming off a year in which ABS CDO spread volatility was relatively low, 2007 is gearing up to be a year in which tiering emerges, fundamentals come to the forefront, and spreads soften further.
PLEASE SEE IMPORTANT ANALYST CERTIFICATION(S) ON THE BACK COVER OF THIS REPORT.
Lehman Brothers | U.S. Securitized Products Research
OVERVIEW
2006: Strong Year for Securitized Products Rate hikes, housing slowdown, low volatility, tight spreads We had expected the past year to be largely a macro story and were not disappointed. Despite the growing dichotomy between the Fed and the capital markets around the strength of the economy, selling volatility has been a winner. Implied as well as realized volatility dropped steadily over the year, spreads moved to the tight end of the range and liquidity premiums have dropped. Against this backdrop, the Securitized Index was the top performer among the major sectors of the aggregate Index, beating swaps by 67bp through mid-December. State of Affairs with the Housing Market We expect housing to be flat in 2007 There has been a dramatic softening in the housing market over the past few months with national HPA dropping from 12% in 2005 to 3.5% in 3Q06. Although home prices are still appreciating, there are three reasons why our outlook for 2007 is pessimistic. First, given that home prices typically have long cycles and that the recent decline has been rather sharp, we dont think we are close to the bottom yet. Second, the fact that 25% of the MSAs are seeing a price decline points to serious structural weakness. Third, the hot MSAs have seen the highest slowdown but have still reported price gains above the national average. We expect housing to stay flat in 2007. Supply Outlook: Lower Cashouts, Weak Underwriting Cashout refinancing volumes have been resilient Purchase activity has declined by close to 25% over the year but cashout refinancing volumes have been surprisingly resilient. With the median new rate of a cashout refinancing materially higher than the old mortgage, we dont expect this trend to persist. Barring a significant rally in mortgage rates, we expect origination volumes to decline an additional 10%-15% in 2007. The proliferation of non-traditional products continued unabated in 2006 with ARMs/non-amortizing products accounting for close to 45% of originations. The impact of lax underwriting is beginning to show in recent performance. However, till credit spreads widen to reflect that, we expect recent origination trends to persist. Demand Outlook: Same as 2006, Risks to the Downside Overseas demand has exceeded our expectations On the demand front, the picture was better than our expectations with banks and overseas players adding $200 billion and $125 billion in mortgages, respectively. Looking into 2007, the picture looks pretty muddled to us. While bank demand for mortgages should be aided by continued asset growth, the appeal of adding carry through duration has dropped significantly. Similarly, overseas demand for mortgages should be helped by continued deficits as well as substitution of reserve holdings into mortgages. At the same time, the recent currency fluctuations throw some uncertainty. Overall, we think that the positives outweigh the negatives and expect the sector to benefit from sustained strong demand factors.
December 12, 2006
Volatility Outlook: Dont Bet the Ranch At the margin, we would sell volatility in 2007 The returns from short gamma were less than a half of those enjoyed in 2004-05. Looking forward, with little potential demand from convexity players, the technical picture for a short volatility trade looks good. On the flip side, the current premium for selling volatility doesnt look encouraging. There is a fair amount of uncertainty around the direction of the Fedmarkets are pricing in significant eases over the year and the official directive from the Fed is still around inflation risks. If bond prices need to readjust, using 1996 as a benchmark, the potential for spikes in volatility is quite high. On balance, we will yield to greed for now and go with a core short volatility strategy. That said, we recommend only a modest position in this trade. From our standpoint, the risk/reward is more in favor of selling liquidity through core longs in CMBS and hybrids. Prepayment Outlook: Value in Call Protection Turnover should slow another 2%CPR in 2007 The impact of HPA on prepayments has been moderate to significant, so far. In prime mortgages, discount prepayments have declined moderately by 2%-3% CPR in aggregate, but have dropped more significantly in the hot MSAs, which have seen the highest slowdown in HPA. Looking into 2007, we expect prime discount prepayments to decline an additional 2% CPR in a flat(tish) housing market. We think the more interesting opportunities on the prepayment front are in call protection stories. Due to a combination of focus on the macro picture and benign refinancings in the recent past, the markets seem to be ignoring call risk. That is especially true for collateral with prepayment penalties. We believe that even a modest rally in rates will bring these refi protection stories to the forefront and look attractive at current levels. Credit Outlook: How Bad is Bad? Worse underwriting and the soft housing market dont bode well for mortgage credit As reflected in our ominous title, our mortgage credit outlook for 2007 is quite bearish. There are two troubling factors, in our mind. First, underwriting standards have deteriorated sharply over the recent past with 06 originations looking about 50% worse than their 03/04 counterparts from a loss standpoint. Second, a soft housing market may exacerbate losses further as these borrowers are unlikely to be bailed out by rising home prices. We expect cumulative defaults on 06 vintages to top 30% if housing stays flat for a year and grows at 5% thereafter. While most of these defaults are set to occur over a longer period, the expected defaults over the coming year are still significant. Going for a Core Short in Mortgage Credit for 2007 The recent spike in delinquencies and potential rating agency action make us bearish We like to be short mortgage credit for three reasons. First, unlike in the past when we were concerned about losses down the road, this time around we are seeing significant delinquencies/losses materialize. Second, with the rating agencies getting very proactive around downgrades, we estimate that a significant proportion of sub-prime BBBs are exposed to negative action over the coming months (Figure 14). This could be the type of headline risk that softens the CDO bid. Third, the growth in the synthetics market has been a significant development in ABS and has the potential to affect spreads. Our valuations suggest generic BBB spreads should be close to 350bp and BBB- around 550bp. Portfolio Positioning for 2007 Modest underweight to mortgages; move up-inborrower credit We are initiating a modest short position in mortgages. While the demand and volatility picture looks mixed, historically tight spreads to swaps leave us bearish on the basis. Our preferred trade is to own a combination of CMBS and hybrids against 30-year fixed rate TBAs. In mortgage credit, we favor moving up-in-borrower-credit. Our recommended trade is to buy alt-A and option ARM subordinates versus their sub-prime counterparts.
3
December 12, 2006
CROSS SECTOR SUMMARY RECOMMENDATIONS

MBS FLOW PRODUCTS Basis Intra-sector Modest underweight to 30-year 5.5s vs a mix of hybrids and CMBS. Overweight 15-year versus 30-year. Buy the FN 5.5 fly. Create synthetic premium substitutes for TBA 6s using 5.5s and 5.5% IOs. Buy DW 5.5/FN6.0 swap convexity hedged. Overweight GN2 premiums through the GN 2/FN 6 swaps to source call protection in a slowing housing market. Overweight LLB FN 6s, new WALA GD 6.5s and 2004 vintage FN 5s vs. TBAs.
PRIME MBS Basis Overweight hybrids versus fixed-rates and non-agencies over TBAs. AAA mezzanine MTA floaters stand out as the most attractive cash product. Source call protection through penalty pools in hybrids, conforming alt-As in fixed-rates, and premium California pools. We favor moving up-in-borrower-credit, down-in-rating. Own alt-A BBs and buy protection on subprime BBBs. Be wary of high CLTV purchase loans; avoid credit exposure to alt-B pools. In option ARMs, we recommend owning subordinates off XS/OC structures.
Intra-sector
Credit
NON PRIME MBS Sub-prime Overweight 1-3 year AAA Floaters versus Consumer ABS due to the 5-15bp carry pickup with minimal AFC risk. 1H05/2H05 subordinates look attractive versus the 2004/2006 vintages after factoring in differences in underwriting, built-up equity and structure. Favor cashout heavy deals versus purchase & full doc versus stated doc due to higher subordination than historical experience. Box trade: Long 06-2 BBB/06-1 BBB- paired with short 06-2 BBB-/06-1 BBB, to express a view on capital structure mispricing regardless of HPA. Move up-in-credit; increased focus on fundamentals; tiering to emerge.
Synthetics
ABS CDOs
December 12, 2006
FLOW PRODUCTS
2006: Strong Spread Product Demand Strong demand from overseas and banks in 2006 Agency mortgages are set to close the year among the top performing spread sectors, having outperformed Treasuries by more than 120bp. Roughly a third of the returns were due to the decline in implied volatility and another third due to the tightening in swap spreads (Figure 1). The demand for mortgages over the course of the year generally exceeded our expectations. The first quarter saw mortgages benefit from corporate crossover flows and strong bank demand. The overall inflation fear-led sell-off in the second quarter saw mortgages underperform their benchmark curves. The rally in rates that started in mid-July on concerns of slower growth saw mortgages lead the way in spread tightening., spurred by servicers adding duration into the rally, robust overseas demand, and bank demand that remained strong in whole loans. The slowdown in housing started to take a toll on discount prepayments, especially in the second half of the year. Turnover rates on 50bp discounts decreased by around 2% CPR over the year. The market largely priced the effects of a slower housing market into valuations of both Trust IOs and up-in-coupons. The most dramatic change in valuations over the year was the tightening of premiums, with most OAS models penalizing the TBA premiums for their higher loan sizes and lower SATOs. The GN and 15-year sectors also lagged, as most of the tightening was led by the 30-year stack. In the IO market, most of the discount IOs held up well adjusted for the flattening of the curve (Figure 2); premium alt-A like IOs such as FNT-361 widened significantly over the year. 2007: Demand Should Stay Strong Prefer selling liquidity over volatility We do not expect any major changes on the demand front in 2007. If the low realized volatility environment holds, mortgages should continue to clip a reasonable amount of carry. We are worried about selling volatility too aggressively, though, given that current market pricing of short rates does not seem to be in line with the inflation concerns of the Fed. At the margin, we would prefer selling liquidity to selling volatility. In the next few sections, we discuss our outlook for mortgage supply/demand, prepayment trends, and mortgage convexity risk/volatility. Our key trade recommendations for the coming year: A modest underweight to 30-year mortgages versus hybrids + CMBS Large overweight to 15-year mortgages versus 30-years Sourcing call protection through GN premiums, LLB collateral, and new WALA 6.5s
Figure 2 Trust IO Returns vs. Collateral
Trust Q1 Q2 Q3 YTD(Nov)
Figure 1 Attribution of MBS Index Returns

Return Attribution vs Swaps (bp) Year 2002 2003 2004 2005 2006 YTD (thru Nov)
Source: Lehman Brothers
Excess Returns 72 (34) 121 (23) 67
Carry + Convexity Losses 70 (59) 38 0 2
Implie d Vols 25 5 59 9 30
Spreads FHT 232 - 2005 5.0% (21) 21 24 (32) 35 FHT 232 - 2005 5.0% FHT 231 - 2005 5.5% FNT 361 - 2005 6.0%
10 (29) (112) 2 2 2
(10) (12) (77) (11) (12) (14)
49 25 (140) (27) (33) (35)
8 (72) (438) (84) (92) (84)
FHT 231 - 2005 5.5% FNT 361 - 2005 6.0%
Performance Attributed to Curve (bp)
December 12, 2006
MBS FLOW PRODUCTS: SUPPLY SIDE EQUATION Agency fixed issuance returns After two years of muted growth, agency fixed issuance surged. Net issuance through October of 2006 has totalled close to $225 billion and accounts for more than 40% of the net increase in all mortgage debt outstanding. In contrast, issuance in 2004 was $-34 billion and 2005 was $90 billion. This can be attributed to two sources: new borrowers and larger loans taken out on the same properties. The net issuance from new borrowers is from new home sales or from existing borrowers refinancing away from other products such as hybrid ARMs to agency fixed rates. The net issuance owing to larger loan sizes occurs from borrowers cashing out of their properties through either existing home sales or through cashout refinancings. A proxy for issuance due to new borrowers in fixed rates is the number of loans outstanding in the fixed-rate universe. The total number of loans was on a decline from 3Q 2002 to 3Q 2005 but has increased dramatically since then (Figure 3). To estimate the new issuance being contributed due to larger loan sizes we looked at the ratio of the loan size of originations vs. the loan size of paydowns (Figure 4). In 2006, new originations were around 7% larger than paydowns. In the 2003-2005 period loans being originated were roughly the same size as paydowns. Increase in Borrowers Primary Driver We split up the cumulative net issuance from 2000 and estimate the net issuance due to an increase in the number of loans and that due to other factors (primarily change in loan sizes) in Figure 5. Of the net issuance of close to $225 billion in 2006, we estimate that over 60% is due to an increase in the universe of fixed rate agency borrowers. Expectations of Supply Net issuance will be rate dependent. Pick up in a rally Looking ahead into 2007, the number of borrowers moving into fixed rate should increase with a rally in rates. This will primarily be due to existing hybrid borrowers refinancing into fixed rates. In addition to rates, the strength of the housing market will also be a big driver of net issuance. In a weakening housing market we think the trend of borrowers cashing our equity from their homes should slow. In Figure 6, we summarize
New borrowers in the system has grown
Figure 3 Number of Loans in Fixed Rates Increasing

Millions of Loans 28 28 27 27 26 26 25 Jan-00
Figure 4 Paydowns are Returning with Larger Loansizes

115% Ratio of Originated Loansizes to Paid Down Loansizes 110% 105% 100% 95% 90% Dec-02 Sep-03
May-01
Sep-02
Jan-04
May-05
Oct-06
Jun-04
Mar-05
Dec-05
Oct-06
December 12, 2006
our outlook for net issuance in the sector in 2007. In summary, a softening housing market should reduce net issuance in the sector but a rally in rates could increase issuance because of ARM borrowers refinancing into fixed rates.
Figure 5. Attribution of Net Issuance in Fixed Rates ($ billion)
Year 2002 2003 2004 2005 2006 Net Issuance 225 266 -34 89 224 Due to Borrower Migration -37 -30 -143 -44 137 Other Factors (Change in Loansizes etc.) 262 296 109 132 87
Source: Attribution by Lehman Brothers. Total Net Issuance from EMBS
Figure 6. Net Fixed Rate Supply by Rates and HPA Environment ($ billion)
Rate Scenario (bp) -100 12% HPA 5% HPA 0% HPA
-50 323 246 169
0 289 207 125
50 267 180 92
100 253 160 66
370 298 225
December 12, 2006
MBS FLOW PRODUCTS: DEMAND SIDE TECHNICALS Demand = Overseas + Banks Predominant demand for passthroughs continues to be from overseas and banks With portfolio restrictions on the agencies and with spreads fairly tight, the marginal demand for the sector continued to be from banks and overseas investors. Over the last four years, banks and overseas investors have absorbed close to 80% of the net supply in agency passthroughs (created from both net issuance as well as a shrinkage in agency portfolios) Figure 7. The demand from these institutions was for the most part better than our expectations for 2006. We had expected a slowdown in bank deposit growth rates which did not materialize and we had underestimated the pace of overseas demand. Overseas portfolios grew at an annualized pace of $125 billion compared with $75 billion in expectations.
Figure 7. Major Holders of Agency Passthroughs ($ billion)
Agency Passthroughs 2002 2003 2004 2005 As of Sep 2006 Change from 2002 2814 3080 3046 3134 3332 518 Agency Portfolios 894 968 831 653 640 -254 Banks 613 855 933 899 948 335 Overseas 140 163 222 322 414 274 Others 1167 1094 1060 1260 1330 163
Source: OFHEO, Freddie Mac and Fannie Mae Monthly Volume Summaries, US Treasury, FDIC, Federal Reserve, Lehman Brothers. We estimate overseas demand in 2006 based on data from the Feds flow of funds and using data from the Treasurys annual survey to assess trends.
Overseas Demand Should Stay Strong

Central Bank Demand
Central bank asset reallocation should continue
Foreign central banks are still largely allocated to Treasuries and these portfolios are gradually being substituted with spread assets (Figure 8). In aggregate the share of agency MBS in their portfolios is still very small. The share is currently 4% of their US credit portfolios. In comparison MBS account for close to 40% of the US Aggregate Index. Continued deficits should bring more money to put to work into US assets. Even if deficits reduce, MBS should benefit from continued substitution out of Treasuries.
Other Overseas Institutions
Private overseas institutions have stepped away from treasuries as well
Private overseas investors have also chosen to slow down the growth in their treasury portfolios. The primary reason for this is just the carry. With most plain duration instruments carrying negatively the additional yield from mortgages is all the more attractive today. Private overseas investors have been reluctant to add Treasuries since the third quarter of 2005 while agency portfolios have continued to grow (Figure 9). In the absence of any dramatic steepening in the curve we would expect this dynamic to continue. Overall we expect overseas portfolios to grow at the same pace in 2007 as in 2006, amounting to about $125 billion of agency mortgages.
December 12, 2006
Figure 8. Share of Assets of Foreign Central Banks

Treasury Jun-04 Jun-05 Market 78% 73% 24% Agency MBS 2% 4% 37% Agency Debt 16% 18% 12% ABS+CMBS 2% 1% 6% Corporates 2% 3% 21%
Source: US Treasury, Lehman Brothers
Figure 9.
Treasury Investments of Private Overseas Investors Have Stalled
Grow th Rate (Q1 2004 = Base) 50% 40% 30% 20% 10% 0% 2004Q1 2004Q2 2004Q3 2004Q4 2005Q1 2005Q2 2005Q3 2005Q4 2006Q1 2006Q2 Grow th of Treasury Portf olios Grow th of A gency Securities Portf olios
Source: Federal Reserve Flow of Funds
Bank Demand Lack of Carry Could be a Dampener Demand from banks will be driven by deposit growth rates but shrinking margins is cause for concern We had expected 2006 to see a softening in the growth rate of bank deposits. Deposits continued to grow at around 8% and in line with this, mortgage portfolios grew at about 7% in 2006 (Figure 10). If deposit growth continues to grow at its historical median pace, the demand for mortgages should be within $175 billion to $ 225 billion a year. Our projections for bank demand are based on assuming deposit growth continues strong and continues to drive mortgage portfolio growth. However the lack of yield in buying mortgages worries us that these sensitivities may not hold. In Figure 11, we show the historical difference between mortgage current coupon yields and average cost of funds at banks. In spite of a more than 400bp increase in the fed funds of the second quarter of 2006, the margin from buying mortgages was still close to historical averages of around 3.5%. Two things have happened since then. First, mortgage rates have rallied by close to 60bp. Second, the cost of funds at banks tends to lag moves in short rates. We project that by mid-2007, mortgage current coupon yields will only be 225bp over the cost of funding if rates follow the forwards or only 60% of historical averages. This margin compression may dampen the strength of bank demand.
Figure 10.
Assets Deposits C/I All Mtgs 1-4 Family First Lien MBS
Source: FDIC, Lehman Brothers
Growth Rates of Bank Assets

2002 7% 7% -7% 12% 12% 13% 2003 8% 7% -3% 7% 7% 8% 2004 11% 10% 5% 13% 13% 13% 2005 8% 8% 12% 6% 9% 2% 2006 YTD 2006 Annul. 8% 6% 9% 5% 4% 6% 11% 8% 13% 7% 6% 8%
December 12, 2006
Figure 11.
6.0% 5.0% 4.0% 3.0% 2.0% 1.0%
Margin from Holding Mortgages Is Low & Likely to Decline
0.0% 31-Dec-93
19-Apr-97 06-Aug-00 Spread Along Forw ards Spread w ith Fed on Hold
24-Nov-03
13-Mar-07
30-Jun-10
Historical Spread Betw een Mortgage Y ields and Cost of Funds

Source: Lehman Brothers, FDIC
Real Money Investors Corporate Crossover Theme Likely to Continue Corporate crossover flows likely to continue in 2007 One of the biggest sources of demand for mortgages in the first quarter of 2006 was from real money investors substituting corporates for discount mortgages. In Figure 12, we show the spread between the ZV spread of 100bp discount mortgages adjusted for the slope of the curve vs. the spread of 5-year A corporates to Treasuries. The spread has tightened marginally over the course of 2006 but still looks attractive from a historical perspective. Mortgages could therefore continue to see demand from corporate investors worried about fundamentals in the credit world.
Figure 12.
bp 60 30 0 -30 -60 -90 -120 1994
Corporate Spreads vs. Mortgages
1996
1998
2000
2002
2004
2006
Curve A djusted ZV of 100bp Discount vs. 5yr A Corp Sprd to Tsy

Source: Lehman Brothers. Curve adjustment is based on a regression of ZV spreads to the 2s-10s slope.
Convexity Risk Unlikely to Help Spreads The duration rebalancing needs of mortgage servicers has been a positive for mortgage spreads in the rally over the last three months. In Figure 13, we compare the duration profile of the MSR universe today vs. mid-July. Before the rally in rates, the risks of servicing portfolios were heavily skewed to the call side. Today the profile is a lot more symmetric in local moves in rates. In July there was clearly more risk of tighter spreads led by servicer receiving in a rally compared to widening in a selloff. Given the change in profiles of the servicing community we do not think convexity risk is a positive for mortgage spreads going forward.
December 12, 2006 10
Mortgage Demand - Spread Volatility Could Increase With demand predominantly from macro investors, spread volatility will increase Overseas demand should continue strong for the sector. On the domestic bank front the lack of carry in the curve is cause for concern but we think finally demand will be led by deposit growth rates. 2007 could also see a continued asset reallocation trade by domestic real money investors from corporates to mortgages. While the core demand for the sector will be strong, given that the primary demand is coming from macro investors and not relative value players, mortgage spreads are likely to be more volatile going forward.
Figure 13.
200 150 100 50 0 -50 -100 -150 1 2 3 4 5 6 7 8 9 10 11
Convexity Profile of the MSR Univ ($B 10yrs)
Prof ile of MSR as of Dec 8th

Source: Lehman Brothers Market Monitor (LehmanLive)
Prof ile of MSR as of July 14
December 12, 2006
11
PREPAYMENTS TRENDS Market is Priced to a Slow Housing Market Turnover rates have slowed; market is priced to slow housing environment Moving to the fundamentals of the sector, the impact of HPA on prepayments has been moderate so far. In Figure 14, we compare the seasoning curves on 50bp discount 30yr mortgages in the second half of 2005 and 2006. In general, prepayments have reduced by about 2% CPR in the 12-24 WALA range. If the housing market stabilizes here we would expect speeds to decline an additional 2-3% CPR over the course of the year. We estimated what the market is pricing in for turnover based on the prices of discount IOs and POs. We modified our model prepayment projections to get similar valuations on the IO and PO. The market implied prepayment curve which is based on this modified model, is significantly slower than the recent experience. While the market seems to have priced in a fairly slow turnover environment for fixed rates, we think valuations are discounting the call risk in the market. Higher coupon valuations on our OAS model are fairly tight with respect to the discounts. The primary reason for this is the higher callability of the high loan size and good credit TBA premiums. Our prepayment themes for 2007 will therefore be focused on sourcing cheap sources of call protection as the impact of a slower housing market on discount prepayments seems fairly priced currently.
Figure 14.
16 14 12 10 8 6 4 2 0 0 6 2006H2 12 2005H2 18 24 30 Market Implied 36
Call protection is being discounted
Seasoning Curves of 50bp Discounts (1m % CPR)
Historical at 5% HPA
Source: Lehman Brothers. Seasoning curves are day count and seasonality adjusted. Historical curves at 5% HPA are based on state level prepayments from Jan 2004 to Oct 2006. The market implied curve uses our prepayment model as a base and adjusts projections to achieve even OAS pricing on a set of Trust IO/POs.
December 12, 2006
12
VOLATILITY OUTLOOK 2007: DONT BET THE RANCH A Moderate Year for Selling Volatility While positive, returns from a short gamma trade were significantly lower in 06 Against a backdrop of a search for assets, it is no surprise that sellers of volatility were plenty in 2006. Longer dated volatility (2y5y, 3y10y) has declined close to 3 vega over the year. Similarly, realized volatility was also very tame. Over the year, realized volatility on 5y and 10y swap rates remained well below 75bp/y with a few exceptions. That said, the premium for selling convexity has also been fairly modest this year (Figure 15). As a result, returns from a short gamma trade in 2006 were between 1/2 to 1/3 of those enjoyed in 2004 and 2005. Will 2007 be Better or Worse? The macro picture and decline in implied vol. make us nervous around shorting volatility in 07 Will 2007 remain another year of low realized volatility and declining implied volatility? With little in the way of demand from mortgage convexity players and enough capital in the system, the technical picture for a short volatility trade looks good. On the flip side, the current premium for selling volatility doesnt look encouraging. Implied volatility for longer dated swaptions is close to all time lows. Short dated options are also pricing in expectations of low realized volatility, the recent modest uptick notwithstanding. Overall, the macro picture really concerns us on selling volatility. There is a fair amount of uncertainty around the direction of the Fedmarkets are pricing in significant eases over the year and the official directive from the Fed is still around inflation risks. It remains to be seen if the Fed follows the market or if bond prices will have to readjust. To the extent it is the latter, using 1996 as a benchmark, the potential for spikes in volatility is quite high (Figure 16). On balance, we will yield to greed for now and go with a core short volatility strategy. That said, we recommend only a modest position in this trade. From our standpoint, the risk/reward is more in favor of selling liquidity through core longs in CMBS and hybrid asset classes.
Figure 15. Returns from Short Gamma Strategy

160 120 80 40 0 -40 -80 1998 1999 2000 2001 2002 2003 2004 2005 2006 Returns on Trade A verage
Figure 16. The 1996 Experience

160 120 80 40 0 01-96 6.5 6 5.5 5 4.5 12-96
03-96 05-96 08-96 10-96 USD 3M 5Y Realised BP V ol (Lef t) Predicted 3m LIBOR(6 months ago) Realized 3m LIBOR
Source: Shows returns on selling 1mx5yr straddles every day and closing out the position at the end of the next day and reopen a new position. Transaction costs are ignored. 2006 numbers are annualized.
Source: LehmanLive. Predicted 3M LIBOR is from the second ED contract.
December 12, 2006
13
RECOMMENDED POSITIONING FOR 2007 Mortgage Basis Spreads, Vols and Prepays The supply/demand picture for mortgages is generally positive. However, given that demand is primarily from macro investors, the potential for spread volatility is high. If the low realized volatility environment holds, the returns from owning mortgages are fairly attractive (Figure 17). We are reluctant to be short carry over the next year but given the tight level of mortgage spreads we think investors are better compensated in adding to other securitized sectors such as hybrids and CMBS. Our recommendation for the mortgage basis is to create substitutes for fixed rate mortgages using hybrids and CMBS. The trade loses a modest 2+/32nds in carry a year (Figure 18). The trade would provide protection in any extreme rate event. In our levered portfolio we are allocating a fairly modest sized position to this trade with $10 million in equity. Technicals for mortgages has tended to be strong in the first quarter of the year; we will look to add to the position over time. In our portfolio we look to make back some of the carry lost in this trade by overweighting hybrids vs. swaps.
Trade - Modest Underweight to Fixed Rates: Sell FN 5.5s versus Jumbo 5.5% 5/1 Hybrids + 10-year CMBS. Figure 17.
-100 -110
Excess Returns on FN 5.5s vs. Swaps 1yr Horizon (bp)

Rate Shift (bp) -75 -36 -50 20 -25 51 0 63 25 54 50 30 75 -4 100 -47
All rate changes are vs. spot rates. The returns are shown over a 1yr horizon vs. swap hedges assuming constant OASs.
Figure 18.
Security FNCL 5.5
Carry on FN 5.5s vs. Hybrids + CMBS Combination

Carry (32nds) 9+/32nds Notional Used -100 63 28
11/32nds (25bp Coupon over 1yr Swaps + 3/32nds Jumbo 5/1 5.5 positive roll down from seasoning) 10yr CMBS Total 0/32nds (25bp over 10yr swaps, flat coupon to 1yr swaps) -2+/32nds carry per year
Up In Coupon Prices vs. OASs In Figure 19, we show the price differential between 50bp premiums and current coupons through 2006 and plot this versus OAS differentials. The price differentials show a gradual cheapening of coupon swaps while OAS differentials tell the opposite story. The reasons for underperformance on a price basis are due to a flatter curve, lower vols and lower spread durations on the higher coupons. The OAS differentials have tightened primarily because of the TBA deliverables in premiums becoming worse over the course of the year. The change in deliverables on FN 6s is worth about 26bp in OAS and this explains the relative tightening of 20bp vs. FN 5.5s (Figure 20). Given that the tightening in OAS almost lines up with the effect of changes in deliverables on the coupon, we argue that the market has ignored the worse quality deliverables in premiums. We recommend being long lower coupons on a curve hedged basis and specifically recommend selling the FN 6.0/5.5 swap.
December 12, 2006
14
Figure 19.
Coupon Swaps Have Cheapened on a Price Basis Rich on OAS?

LOAS (bp) 10 5 0 -5 -10 -15 -20 03-06 05-06 07-06 09-06 -25 12-06
32nds (Sw ap Price) 15 10 5 0 -5 -10 -15 12-05
Price of 50bp Premium vs. Current Coupon (Left) LOAS of 50bp Premium vs. Current Coupon (Right)
Figure 20.
Security 30yr FN 4.5 30yr FN 5.0 30yr FN 5.5 30yr FN 6.0 30yr FN 6.5
Effect of Changes of Deliverables on TBA OASs

OAS with Current Deliverables -5.5 -11.9 -16.6 -32.1 -36.6 OAS with Same Deliverables as Dec 2005 -13.7 -14.5 -10.9 -6.3 2.4 Difference 8 3 -6 -26 -39
Issuance Should Cheapen FN 5 Deliverable Lower rates today compared to earlier in the year should cause increased issuance in the lower coupons. We estimate that by February of 2007, there should be an additional $10 billion in new FN 5s being securitized. These pools originated in a weak housing market are likely to season fairly slowly and to lower fully seasoned rates than in the past. We recommend positioning for a cheapening in the FN 5 deliverable by being long FN 5.5/5.0 swap. Putting our two recommended coupon swap trades together, we recommend buying the FN 5.5 fly.
Trade Up in Coupon is Rich, FN 5s Could Cheapen: Buy FN 5.5 fly
IOs vs. Coupon Swaps IO Market is Cheaper In Figure 21, we show a regression residual of the prices of synthetic 6s created off of TBA 5.5s and FHT-237 IOs versus prices of TBA 6s. Based on this regression, we think the synthetic security looks about 3/32nds cheap. While premium securities have richened over the past month, IOs have not kept pace. In the levered portfolio, we recommend buying 100% FN 5.5s with 9% FHT-240 IOs hedged with 115 TBA 6s and are looking for an upside of 3/32nds.
Trade IOs are Cheap to Up in Coupons: Buy TBA FN 5.5 + 5.5% IO and Sell TBA 6s.
December 12, 2006
15
Figure 21
4/32 3/32 1/32 0/32 - 1/32 - 3/32 - 4/32 08/06
Regression Residual of Prices of Synthetic 6s vs. TBA 6s
09/06
10/06
11/06
12/06
Long 15-year vs. 30-year Convexity Hedged The tightening of the mortgage basis since mid-July has been led by the 30-year stack. The relative spreads between the 15-year and 30-year sectors have widened and the 15year picks up close to 10bp in OAS versus its 30-year counterpart. We do not recommend that investors hold on to the long gamma and negative carry positions in owning 15-year vs. 30-year one for one. We instead recommend buying 15-year and selling enough 30-year to be convexity neutral and offsetting the remainder of the curve and duration risk using swaps. We recommend the trade through the DW 5.5/FN 6.0 swap in our levered portfolio.
Trade 15yrs Attractive to 30yrs: Buy DW 5.5s, Sell FN 6s and Hedge Residual Duration and Curve Exposure with Swaps
GN/FN Swaps Sourcing Call Protection GN premiums afford two sources of call protection versus conventionals. First, given the weaker credit nature of the GN borrower, their refinancing behaviour is more sensitive to a slowdown in the housing market. Second, lower loan sizes on GN TBAs also provide incremental call protection. Current payups of 12/32nds on the GN2/FN 6 swap, are not too far from historical averages. Since late last year, the TBA quality of FN 6s has deteriorated significantly because of higher loan sizes. We think the fair payup for just the difference in loan sizes should be of the order of 12/32nds. At current market payups the explicit guarantee of GNMAs, the lower delay on payments as well as the weaker credit nature of these pools can all be obtained for free. We are looking for 7-8/32nds of repricing on the swap.
Trade Overweight GN Premiums: Buy GN2/FN 6 Swap
Specified Pools Buy LLB FN 6s: We retain our current recommendation in our levered portfolio to source call protection through LLB collateral. While payups on this collateral have increased fairly significantly over the last couple of weeks, with LLB 6s trading up close to half a point to TBAs, we think the collateral is still trading at under 50% of fair value. We recommend hedging the position to 110% of TBA durations.
December 12, 2006
16
Buy New WALA GD 6.5s: Given the rally in rates any new production in 6.5s will be high SATO pools and are likely to be from credit impaired borrowers. Given weakening housing fundamentals the credit curing in this collateral will be extended. The payups on new WALA FN 6.5s are very nominal at around 1/32nds. We recommend buying new WALA FN 6.5s and selling TBAs against it. Buy 2004 5s vs. TBAs: We expect total new production of about $10 billion in FN 5s by February. This new production has been originated in a fairly weak housing market and is likely to season much slower than the older vintages. 2004 vintage 5s are currently trading at a payup of 6/32nds to TBA. If the TBA does cheapen to a 2 WALA deliverable by February, we think the payups on the 2004 vintage should go up by 34/32nds.
Specified Pool Trades 1. Buy LLB FN 6s vs. TBAs 2. Buy new production GD 6.5s vs. TBA GDs 3. Buy 2004 FN 5s vs. TBAs
December 12, 2006
17
LEVERAGED PORTFOLIO Our leveraged portfolio that tracks our trade recommendation is up 7.8% ROE for the year, versus 10% 2004 and 2005 and 18% in 2003. We summarize our best and worst five trades over the last year in Figure 22. We also show the cumulative performance of the levered portfolio since its inception in 2003 (Figure 23). Figure 24, shows the current positions in our portfolio that we initiated based on recommendation in our 2007 outlook.
Figure 22 Our Best and Worst Calls for the Year
Figure 23 Cumulative ROE of Leveraged Portfolio

45
Profit* 1,930 1,593 1,476 1,191 852 -1,356 -1,255 -394 -386 -341
Best & Worst Retired Trades - Last 12 Months

Trade 1 Buy MTA BBBs and Buy Protection on BBB HELs 2005 Vintage 5.5% IO + 2003 Vintage 5.5% PO vs. TBA 5.5s Holding Period 8/21/06 - 12/11/06 4/17/06 - 10/20/06 12/15/05 - 2/23/06
40 35 30 25 20 15 10 5 0 01/03 01/04 12/04 12/05
2 Buy DW 6 vs. 2yr/10yr Swaps 3
4 Buy FHT-231 IO vs. Collateral, Curve Hedged 12/15/05 - 8/4/06 5 AAA Alt-A IO hedged with 10yr Swaps 1 Buy FN 5.5/6.0 Swap 2 Buy FN 6/5 Swap with 10yr Swaps 3 Sell Mortgages vs. Hybrids/CMBS 4 Buy DW 5/4.5 Swap 5 Buy DW 5/FN 5 Swap with 10yr Swaps *($, 000s) 4/21/06 - 8/4/06 9/25/06 - 12/11/06 12/15/05 - 2/23/06 12/15/05 - 4/13/06 6/12/06 - 7/21/06 12/15/05 - 1/20/06
SUMMARY POSITIONS IN LEVERAGED PORTFOLIO Long

1 2 3 4 Sell Mortgages vs. Hybrids/CMBS Buy FN 5.5 Fly Synthetic Premium 6s vs. TBAs Overweight DW 5.5/FN 6.0 swap convexity hedged Buy GN2/FN 6 Swap Buy LLB 6s vs. TBA Buy New Production GD 6.5s vs. TBA Buy 2004 FN 5s vs. TBA Jumbo 5/1 5.5 vs. Swaps Jumbo 6s vs. TBA 6s $63 FN 5/1 5.5% $28 10yr CMBS $200 FN 5.5 $100 FN 5.5 $9 FNT-237 IO $100 DW 5.5
Short
$100 TBA FN 5.5 $100 FN 5.0 $100 FN 6.0 $115 FN 6 $78 FN 6 TBA 2yr/10yr Swaps to Curve Hedge Position $100 FN 6s $110 TBA FN 6 $100 GD 6.5s $100 TBA FN 5s 2yr/10yr Swaps
Equity
10 15 10 20
Leverage
10 25 20 25
Comments
Modest underweight to the basis vs. other securitized sectors. Both wings of the fly look rich. IO valuations cheaper than premiums Overweight 15yrs but not onefor-one Buy GN premiums as a source of call protection LLB collateral for call protection New production in 6.5s is high SATO, will cure slower in a slower housing marker Production in FN 5s should boost valuations of seasoned collateral Overweight basis through hybrids Overweight Non-Agencies MTA Credit vs. Sub-Prime
5 6 7 8 9 10
$100 GN2 6s $100 LLB FN 6 $100 1 WALA GD 6.5 $100 2004 FN 5s $100 Jumbo 5/1 5.5%
10 10 10 10 30 10 15
20 20 20 20 20 20 5
$100 Jumbo 6% Fixed $100 TBA FN 6% $100 MTA BBB & MTA BBBs vs. Sub-Prime Protection on Sub11 BBB Prime BBB All Trades will be added to portfolio at levels as of close of 12/12/2006
December 12, 2006
18
PREPAYMENT OUTLOOK
PREPAYMENTS IN A CHANGING HPA ENVIRONMENT At 5% HPA and with unchanged rates, we expect 30-year fixed-rate paydowns to increase by 44% in 2007 relative to 2006. Flat HPA would lower speeds by 10%. Bad credits (high SATOs and GNs) and new originations would be most affected by an HPA slowdown.
Last year we discussed the possible effects of HPA on prepayments, cautioning our readers that most of the action would have taken place in 2007. While discount speeds in the past few months felt the pinch of the lower HPA, we still think most of the effects will take place next year, as new cohorts get originated in a low HPA environment. The main topic of this year therefore remains HPA; we delve deeper into how different collateral could be affected by this new environment. 2006THE TURNING POINT FOR HPA Speeds had a gradual slowdown for most of 2006 With the exception of recent premium speeds, prepayments in 2006 went through a gradual slowdown across the board. Premiums were in general relatively slow, but those cohorts were either bad credits or older, burnt-out originations. (For more details, please refer to the appendix on 2006 prepayments of major cohorts in the various sectors.) Later, we consider the extent to which new cohorts could be callable in a rally. Most of the action took place in the discount space. As home prices hit the brakes, purchase activity also declined to levels last seen in 2000 (Figure 1). Although the summer seasonals hid the slowdown in discount speeds, the seasonally adjusted speeds in Figure 2 clearly show how discounts gradually slowed since about March (see the Appendix for other sectors). FN 4.5s of 03 slowed by more than 20% Even FN 4.5s of 03 slowed significantly despite leading the fast discount speeds over the past couple of years. The cohort came in at a more reasonable 7.3% CPR for the 12 months ending in November, a far cry from the 9.9% CPR of the previous 12 months. Although a little over half of the slowdown was due to rates, almost a full CPR was
Figure 2. and Discount Speeds Slow
CPR CPR 14 12 15 10 5 0 1998 10 8 6 4 2 2000 US HPA 2002 2004 2006 EHS CPR 0 Dec-05 Feb-06 Apr-06 Jun-06 5.0 05 Aug-06 5.0 04 Oct-06 5.0 03
Figure 1. HPA and Purchase Activity Decline

HPA 20
MBA Purch Index CPR
4.5 03
Source: Lehman Brothers, FHLM, NAR, Federal Reserve Flow of Funds, MBA
Source: FNMA, Lehman Brothers. FN aggregates only.
December 12, 2006
19
likely HPA-related, a reasonably large proportion for a cohort with about 45% accumulated HPA. Recent speeds were even lower; on a daycount and seasonally adjusted basis, speeds in the past three months were 6.5% CPR, down from 8.5% CPR at the beginning of the year. Hybrids had a similar percentage slowdown and so did GNs, even if they faced a smaller HPA slowdown Hybrids (Figure A10 in the Appendix) slowed similarly in percentage terms, with 5/1 4.5 of 04 printing an adjusted 18.7% CPR over the past three months, versus 24.7% CPR at the beginning of the year. Finally, GNMA seasoned cohorts also exhibited a similar percentage slowdown (Figure A7 in the appendix), but GN cohorts are concentrated in states with lower HPA (e.g., GN 4.5 03 have about 85% of the HPA of their corresponding FN and FH counterparts), pointing to a higher sensitivity to HPA.
Figure 3. High Share of Refi in a Backup

100 9.00
Figure 4. As Purchase Slows While Refis Are Stable
75
8.00
50
7.00
25
6.00
0 1998
5.00 2000 2002 REFI % 2004 2006 1998 2000 2002 2004 2006
30y Rate Blended
CPR Refi Approx
CPR Purchase Approx
Source: Lehman Brothers and MBA.
Source: Lehman Brothers, MBA, Federal Reserve. We take the dollar volume of purchase and refinancing applications and divide by the size of the mortgage universe. The results do not show a scale because the MBA index only represents the change in applications and not the absolute value.
Figure 5.
Cashouts Are Still High
100% 80% 60% 40% 20% 1993
40% 30% 20% 10% 0% 2006
1995
1998 Cashouts as % of Refi
2001
2004
Average Cashout %
Source: .FHLM.
December 12, 2006
20
More HPA-driven Slowdown? HPA can slow speeds further . as cashouts are still high In the absence of a rate rally, there is definitely more room for an HPA-driven slowdown, although this would be most pronounced in the lower credits rather than in the core of the traditional agency fixed-rate products. Refinancings, as reported by the MBA, are at an exceptionally high level for a backup. The share of refinancings among all mortgage applications (Figure 3) has been in the 40%-50% range for the past couple of years even as rates kept going up. As a comparison, it was below 25% during the 2000 backup. This change was driven by an increasing amount of cash-out refinancing (Figure 5) and by changes in the composition of the mortgage universe, which is now more skewed away from traditional agency fixed-rate mortgages into hybrids and high LTV loans. Both of those drivers increase the sensitivity of overall refinancings to HPA, albeit there appears to be a delayed response. While discounts have already started adjusting, credit impaired borrowers, mostly with a high LTV, will likely slow substantially if HPA grinds to a halt. High LTV Borrowers Could Lead the HPA-driven Slowdown Bad credits are most affected by a HPA slowdown, especially GNs and high LTVs While worse credit loans are less responsive to rate changes, they are more responsive to HPA (Figure 6). This is even more the case for GNMA collateral (Figure 7) and to some extent for high LTV FN cohorts (Figure 8). Home prices affect credit impaired borrowers directly in two related ways. First, they lower their LTV, improving the rate they could refinance into. Second, they build up equity in their homes. This additional equity can be tapped into for a cash-out refinancing, or may be used to trade up to a larger home (albeit this requires a larger mortgage). Fast home price appreciation also affects bad credits indirectly, as it is often correlated with increased local economic activity, which allows them to improve their income and credit. The overall effect can be substantial. Even 50bp in the money, borrowers with over 100bp of SATO have been significantly faster than less credit impaired borrowers in areas where home prices have risen at more than 20% per year.
Figure 6. Bad Credits (High SATOs) Are More Responsive to HPA 0bp Relative Coupon
CPR 35%
50bp Relative Coupon
CPR 45%
25%
35%
15%
25%
5% 5% 10% 15% HPA SATO: -40 0 40 80 20% 25% 30%
15% 5% 10% 15% HPA SATO: 0 40 80 120 20% 25%
Source: FNMA, FHLM, Lehman Brothers. 30-year FRM collateral, WALA 12-18, September 2003 June 2006.
December 12, 2006
21
Figure 7. GN Discounts Are Very Responsive to HPA

CPR, 3m 40% 30% 20% 10% 0% 0 6 12 WALA HPA: 5% 10% 15% 20% 25% 18 24 30
Figure 8. And So Are High LTV FN

CPR, 3m 40% 30% 20% 10% 0% 0 6 12 WALA HPA: 5% 10% 15% 20% 25% 18 24 30
Source: GNMA, Lehman Brothers, FHLM. 50bp out of the money collateral, Sep 03 Jun 06.
Source: Lehman Brothers, FHLM, FNMA. 50bp out of the money collateral with LTV>80%, Sep 03 Jun 06.
Figure 9. HPA Impact on Agency Fixed-Rate Discounts

CPR, 3m 20% 15% 10% 5% 0% 0 6 12 18 WALA HPA: 5% 10% 15% 20% 25% 24 30 36
Figure 10. And on Conforming Alt-A 5/1s

CPR, 6m 50% 40% 30% 20% 10% 0% 0 6 12 18 WALA 5% 10% 15% 20% 25% 24 30 36
Source: FNMA, FHLM, Lehman Brothers. All 30-year non-penalty non-relo FRM conventional collateral. September 2003-June 2006, 50bp discounts.
Source: LoanPerformance, Lehman Brothers. All 5/1 non-penalty alt-A collateral. September 2003-June 2006, 50bp discounts.
Moderately Seasoned Collateral Most Affected Among Discounts Recent cohorts originated in a low HPA will slow the most HPA has already started having a considerable impact on discounts, but the slowdown could be much more pronounced. While older vintages such as 2003 have already been significantly affected (FN 4.5 of 03 is now printing 6.5% CPR) and will likely continue to slow gradually, the effect would be the largest on new production. Currently fixed-rate 50bp discounts print 9% CPR and 5/1s 13% CPR at 12-15 WALA. If HPA was zero from now on, speeds on those moderately seasoned discounts would be down 40% by early 2008. As we show in Figures 9 and 10, the fast HPA of the past few years not only increased the long run turnover of discount cohorts, but also the rate of convergence to this long run level. Discounts in areas with more 20% HPA reached their fully seasoned turnover already at around 18-24 WALA, while it took at least 36 months for those in low HPA areas.
22
December 12, 2006
Figure 11. S-Curve Seems to Have Flattened

CPR 75 60 45 30 15 0 -100
Figure 12. As Justified by Loan Characteristics

Loan Size SATO FICO Aug02-Sep03 Oct03-Oct05 Nov05-Oct06 Nov06-Oct07 -100 160,316 152,388 137,932 209,060 -13 30 116 -1 713 706 665 723 LTV 75.5 76.2 82.6 73.2 NOO 5.5 12.5 21.5 9.6
100 Relative Coupon
200
Aug02-Sep03
Oct03-Oct05
Nov05-Oct06
Source: Lehman Brothers, FNMA, FHLM. 12-24 WALA.
Source: Source: Lehman Brothers, FNMA, FHLM. For historical periods, 12-24 wala, 100bp relative coupon. For Nov06-Oct07 -100, we take pools that are today 0-12 wala, at the money, so that they would be 12-24 wala in the period, and 100bp relative coupon with a 100bp rally. NOO is non-owner-occupied, i.e. non first home.
Has Callability Changed? Recently, premiums have been slow because of credit impairment and lower loan sizes With most attention going to HPA, callability does not seem to be the main concern of investors these days. A cursory analysis suggests that callability seems to have decreased (Figure 11), with recent 100bp in-the-money speeds around 35% CPR, much less than the 56% CPR of 2002-2003. Much of this is obviously driven by changes in the universe of premium mortgages. As rates have been mostly increasing over the past few years, the only way for a mortgage to be 100bp in the money was to have been originated as a high SATO, hence most likely a bad credit. Figure 12 confirms this; recent 12-24 WALA premiums were much lower FICOs, higher LTVs and lower loan sizes than those that faced the 2002-2003 refinancing wave. Should we have a new refinancing wave, though, it is the current coupon new production that would constitute most of the premium universe. The characteristics of this cohort (mostly the larger loan size) make it more callable than the 2003 premiums. Other factors, though, come into play. First, the slope of the curve in 2003 was much steeper. If a rally in 2007 leaves the slope unchanged, prepayments would miss the fixed-to-ARM refinancing component. Second, partly compensating for this, originators have quite a bit of overcapacity, so we would not expect as much widening in the primary to secondary mortgage rate spread this time unless rates rally substantially. Finally, HPA is most likely going to be slower, but we think that home prices have only a small effect on low SATO premiums beyond the effect on turnover. All things considered, we expect callability to be similar to slightly higher than in 2003 unless the rally comes with a steepener, in which case premiums would be faster.
But in a rally, speeds should actually be similar or higher than the 2003 peaks
December 12, 2006
23
Figure 13. 3/1 Reset Effect

CPR 80 60 40 20 0 0 12 24 WALA 3/1 4.5 02 3/1 5 02 3/1 3.5 03 3/1 4 03 36 48
Figure 14. 5/1 Reset Effect

CPR 80 60 40 20 0 0 12 24 36 WALA 5/1 5 01 5/1 5.5 01 5/1 6 01 48 60 72
Source: FHLM, FNMA, Lehman Brothers
Source: FHLM, FNMA, Lehman Brothers
Hybrid Tails Speeds Have Not Been Too Fast Post-reset hybrid speeds have peaked around 70% CPR As hybrid originations season, post-reset hybrid speeds are becoming an important topic, especially for secondary markets. At the same time, we start having data from the first sizeable waves of resets from 3/1s of 2002-2003 and 5/1s of 2001 (Figures 13 and 14). For both cohorts, reset speeds have only temporarily peaked around 70% CPR, fast, but not as fast as most investors feared, especially given the flat slope environment. Two things need to be noted. First, 5/1s of 2001 are hardly an ideal cohort to study the effect of resets. This vintage went through a significant rally in 2003 and is somewhat burnt out. Second, 3/1s of 2002 which have now had two yearly resets, have actually showed an almost identical second spike at the second reset. While the vintage was more burnt out by the second reset, this actually created an even larger shock than the first one. We expect resets in 2007 to be only slightly higher than the 2006 experience. While 5/1s of 2002 are less burnt out and have lower teaser rates than their 2001 counterparts, they will also most likely face a much lower home price appreciation environment. SCENARIOS FOR 2007 With flat rates and 5% HPA, 30-year paydowns should increase by 44% Cashouts should decline during 2007 If rates dont move from current levels, paydowns should, in general, be higher in 2007 than in 2006, even at lower HPAs. With the current 30-year mortgage rate at 6.04%, down from 6.72% in June-July 2006, we expect 30-year paydowns to be up 44% in a 5% HPA scenario, but only 29% with flat HPA. We show projections in flat and up 5% HPA scenarios. While most of the effects of the lower HPA in 2006 should be felt in 2007, similar to what we discussed last year, the effect of these HPA scenarios is not immediate, and the difference between flat and up 5% HPA is larger toward the end of 2007 and in full effect in 2008. Cashouts should slow gradually in both scenarios, although the cashout share of refinancings as reported by Freddie Mac will likely drop more substantially over the next few months as rate refinancing picks up.
2007 resets should only be a little faster
December 12, 2006
24
Still, lower HPA would slow 2006 cohorts by 10%-20% for fixed-rates and 20%-30% for hybrids. For hybrids 5/1 and longer, a flat HPA environment would actually counter most of the effect of lower rates in 2007 relative to 2006. A 100bp rally would take speeds almost back to peak 2003 levels Callability is still high. A 100bp rally would increase fixed-rate paydowns to more than four times last years values, taking them only 5%-10% below the peak of September 2002-August 2003. Our full projections are in Figures 15 and 16.
Figure 15. Fixed Rate Prepayment Projections for 2007 by HPA and Rate Scenario
Historical (Nov 06) Cohort 03 4.5s 06 5s 05 5s 04 5s 03 5s 06 5.5s 05 5.5s 04 5.5s 03 5.5s 06 6s 05 6s 04 6s 03 6s 06 6.5s 02 6.5s 01 6.5s 06 7s 1-Mo 6.0 4.4 7.3 8.3 8.2 5.1 10.4 11.6 11.0 8.9 14.4 15.9 15.1 15.5 15.4 15.4 35.3 18.4 18.7 13.0 17.6 17.3 8.8 12.3 12.3 6.3 8.8 9.3 1-Yr 7.3 -100 bp 15.4 39.7 38.6 33.7 31.3 57.8 50.1 46.4 43.7 64.8 55.7 52.0 44.5 57.5 39.0 35.8 49.2 334% 0% HPA Static* 7.6 6.0 8.4 9.4 9.5 10.3 11.0 11.5 11.6 22.6 19.1 16.6 14.7 32.7 19.0 20.3 37.9 29% +100 bp 6.5 3.9 5.7 6.6 6.7 4.0 6.1 6.9 7.3 4.5 7.4 8.7 9.3 6.8 11.6 12.1 14.1 -38% -100 bp 15.8 40.9 40.0 34.7 32.3 58.8 51.7 47.8 44.9 66.0 57.6 53.7 46.0 59.3 39.9 36.4 52.2 347% 5% HPA Static* 7.8 7.2 9.6 10.1 10.3 11.9 12.9 13.1 12.9 24.8 21.9 18.8 16.7 35.3 20.0 21.1 41.4 44% +100 bp 6.6 4.7 6.5 7.1 7.4 5.1 7.4 8.0 8.3 6.4 9.9 10.6 11.1 9.7 12.7 12.9 17.9 -26%
Paydown**
* Static projection as of December 7, 2006, at an unchanged 6.04% mortgage rate. ** Paydown numbers are the percentage deviation of projected 2007 annual paydown to historical 2006 annual paydown.
December 12, 2006
25
Figure 16. Hybrid ARM Prepayment Projections for 2007 by HPA and Rate Scenarios
Historical (Nov 06) Product 3/1s Coupon 3.5 3.5 4.0 4.5 4.5 5.0 5.0 3/1 Average 5/1s 4.0 4.0 4.5 4.5 5.0 5.0 5.5 6.0 5/1 Average 7/1s 4.5 4.5 4.5 5.0 5.0 5.5 5.5 7/1 Average 10/1s 4.5 5.0 5.0 5.0 5.5 10/1 Average Overall Average Pct Chg from Static 5% HPA
*Static projection based on rates as of December 7, 2006.
0% HPA -100 bp 62 61 56 65 43 54 50 56 36 25 30 29 50 36 43 53 34 23 24 23 31 30 45 50 28 19 28 30 35 54 36 36 47% Static* 55 53 47 54 31 47 35 46 23 17 17 15 40 15 17 25 18 15 13 11 18 12 16 16 14 10 12 10 9 11 10 21 -15% 100 bp 46 44 38 48 22 44 24 38 18 13 13 10 33 10 11 11 13 10 9 7 13 8 10 8 9 7 8 6 6 5 6 15 -37% -100 bp 64 63 59 66 47 56 54 59 39 28 34 33 52 40 47 56 37 25 27 27 33 34 48 53 31 21 30 32 38 57 38 39 61%
5% HPA Static* 57 56 50 57 36 50 41 50 27 21 21 18 43 19 22 31 22 18 16 14 21 16 21 21 17 12 15 12 12 15 13 24 0% 100 bp 49 47 42 51 28 47 31 42 21 15 15 13 36 13 15 16 16 12 11 9 15 10 14 12 11 8 10 8 8 7 8 18 -26%
Orig Year 2003 2004 2004 2002 2005 2002 2005
1-Mo 45.8 28.7 27.8 62.9 23.8 45.5 24.2 32.6
1-Yr 41.7 28.0 27.2 55.3 24.4 50.2 26.2 30.9 20.9 18.0 18.5 14.1 27.8 13.6 11.1 12.7 17.0 17.8 14.7 10.8 20.1 11.1 12.0 7.9 14.1 12.4 15.1 11.5 8.6 5.2 9.3 17.7
2003 2004 2004 2005 2002 2005 2006 2006
18.1 15.4 17.1 14.3 27.7 15.3 13.8 15.5 16.4
2003 2004 2005 2002 2005 2005 2006
15.1 13.4 11.4 17.5 11.4 16.6 8.1 13.2
2003 2003 2004 2005 2006
10.0 15.0 12.9 9.8 6.2 10.2 17.7
December 12, 2006
26
APPENDIX
Figure A1. FN 30 Discounts Slowed with HPA
CPR 14 12 10 8 6 4 2 0 Dec-05 Feb-06 Apr-06 Jun-06 5.0 05 Aug-06 5.0 04 Oct-06 5.0 03 6 Dec-05
5.5 05
Figure A2. Limited Action in FN 30 Premiums

CPR 18 15 12 9 Mtg Rate 6.80 6.60 6.40 6.20 6.00 Oct-06
M tg Rate
Feb-06
Apr-06
Jun-06
5.5 03
Aug-06
6.0 05
4.5 03
5.5 04
Daycount and seasonally adjusted prepayments. Source: Lehman Brothers and FNMA.
Daycount and seasonally adjusted prepayments. Source: Lehman Brothers and FNMA.
Figure A3. Conforming Alt-A Discounts

CPR 14 12 10 8 6 4 2 0 Dec-05 Feb-06 Apr-06 5.0 05 Jun-06 5.0 04 Aug-06 5.0 03 Oct-06
Figure A4. Conforming Alt-A Premiums

CPR 20 15 10 5 0 Dec-05
5.5 05
Mtg Rate 6.80 6.60 6.40 6.20 6.00 Oct-06

M tg Rate
Feb-06
Apr-06
Jun-06
5.5 03
Aug-06
6.0 05
5.5 04
Daycount and seasonally adjusted prepayments. Source: Lehman Brothers and LoanPerformance.
December 12, 2006
27
Figure A5. Non-Conforming Discounts

CPR 14 12 10 8 6 4 2 0 Dec-05 Feb-06 Apr-06 5.0 05 Jun-06 5.0 04 Aug-06 5.0 03 Oct-06
Figure A6. Non-Conforming Premiums

CPR 20 15 10 5 0 Dec-05
5.5 05
Mtg Rate 6.80 6.60 6.40 6.20 6.00 Oct-06

M tg Rate
Feb-06
Apr-06
Jun-06
5.5 03
Aug-06
6.0 05
5.5 04
Figure A7. GN 30y Discounts

CPR 20 16 12 8 4 Dec-05
Figure A8. GN 30y Premiums

CPR 24 19 14 9 4 Dec-05 Feb-06 Apr-06
5.5 05 5.5 04
Mtg Rate 6.80 6.60 6.40 6.20 6.00 Jun-06 Aug-06 Oct-06
5.5 03 6.0 05 M tg Rate
Feb-06
Apr-06
Jun-06 5.0 05
Aug-06 5.0 04
Oct-06 5.0 03
4.5 03
Daycount and seasonally adjusted prepayments. Source: Lehman Brothers and GNMA.
Daycount and seasonally adjusted prepayments. Source: Lehman Brothers and GNMA.
December 12, 2006
28
Figure A9. Agency 3/1s

CPR 35 30 25 20 15 Dec-05 Mtg Rate 6.60 6.35 6.10 5.85 5.60 Oct-06
3/1Rate
Figure A10. Agency 5/1s

CPR 30 25 20 15 10 Dec-05 Mtg Rate 6.60 6.35 6.10 5.85 5.60 Oct-06
5/1Rate
Feb-06
3/14.0 04
Apr-06
Jun-06
Aug-06
3/15.0 05
Feb-06
Apr-06
5/14.5 04
Jun-06
Aug-06
3/14.5 05
5/14.0 03
5/15.0 05
Daycount and seasonally adjusted prepayments. Source: Lehman Brothers, FNMA and FHLM.
Daycount and seasonally adjusted prepayments. Source: Lehman Brothers, FNMA and FHLM.
Figure A11. Conforming Alt-A Neg-am and 3/1s

CPR 60 50 40 30 20 10 0 Dec-05
Figure A12. Conforming Alt-A 5/1s

CPR 40 35 30 25 20 15 10 5 0 Dec-05 5.85 5.60 Oct-06
5/1Rate
Mtg Rate 6.60 6.35 6.10
Feb-06
Apr-06
Jun-06
Aug-06
Oct-06 3/1 4.5 04
Feb-06
Apr-06
5/14.5 04
Jun-06
Aug-06
Negam 04 PPP
Negam 05 PPP
5/14.5 03
5/15.0 05
December 12, 2006
29
PRIME MBS
2006: Another Strong Year Shift away from hybrids into option ARMs and fixed IOs Prime mortgage supply remained robust this past year, matching 2005 levels, although volumes shifted away from hybrids into Option ARMs and fixed-rate IO mortgages, as we expected. In line with other spread sectors, prime AAAs saw significant tightening in 2006, although credit spreads in the alt-A market have lagged the sub-prime market. Prepayments slowed and delinquencies picked up on the heels of a slower housing market, but the effect of this slowdown has not been reflected in performance yet. 2007: All Themes Lead to Housing Expect slower refis and higher defaults in slower housing The strength of the housing market has significant implications for (and is set to be the most important driver of) supply, prepayment and credit performance in 2007. Supply: While aggregate supply should drop in 2007 compared with 2006, the magnitude of the decline could be smaller than expected because of robust cashout activity. Overall, we expect prime supply to be down about 10% in 2007. Neg-Am products are the favorite cashout vehicle and hence, may see robust supply in 2007. Moreover, share of weaker credit loans may remain high despite HPA slowdown. Prepayments: In prime MBS, aggregate turnover has slowed 1-2% CPR while refinancings have dropped 4-5 CPR. Jumbos and option ARMs, which have been largely concentrated in hot MSAs, have seen a bigger drop in speeds than agency/subprime pools, which are more geographically diverse; this is because the HPA slowdown has largely been in these hot MSAs. Moreover, the market is ignoring call protection stories. Credit: Delinquencies on prime originations have surged in recent months, although far less than the increase in sub-prime delinquencies. That said, the uptick in delinquencies is largely a result of the deterioration in underwriting, as the effect of slower HPA is not yet reflected in credit performance. Going forward, we expect alt-A and Option ARM performance to hold up much better in a slower housing market than sub-prime. We are, however, wary of alt-B exposure, where initial stage delinquency trends appear closer to subprime market, and rating agency levels appear inadequate. Own hybrids versus fixed-rates: Hybrids stand to benefit from favorable technicals: we expect lower supply given the drop in purchase volumes, and increased demand from index rebalancing flows following the inclusion of hybrids in the Lehman Brothers U.S. Aggregate Index. Source Call protection: The market is discounting call risk in mortgages; we recommend sourcing call protection through penalty pools in hybrids, weaker credit conforming alt-A fixed-rate pools. Prime credit: Move up-in-credit, down in rating: At current subordination levels, we favor alt-A subordinates over alt-B, option ARMs over sub-prime. In alt-A shifting interest structures, we favor lower rated subordinates such as BBs hedged with CDS on subprime BBB-s.
Key Trade Recommendations
December 12, 2006
30
TRENDS IN SUPPLY/ DEMAND HPA Slowdown Shakes Up Product Mix Purchase volumes dropped significantly as HPA slowed Interest only (IO) loans and pay option ARMs dominated mortgage supply in 2005, as borrowers sought out affordable financing in a steaming housing market. As HPA softened throughout 2006, purchase volumes were hit particularly hard, though cashout refinancings remain high. In turn, neg-am loans, the favored cashout vehicle, now account for over 40% of non-agency supply (Figure 1). Increase in Weaker Credit Originations Further increase in limited documentation loans At an aggregate level, credit quality of recent originations continues to drift lower, with lower FICO borrowers, higher share of loans with piggyback seconds, and perhaps most significantly, a further increase in limited documentation loans. In particular, the alt-B sector has seen the largest increase in leveraged borrowers with limited doc loans. Outlook for Supply in 2007 Cashout refi products such as neg-am loans should stay high Cashout refinancing volumes are likely to be sustained through 2007 on the back of builtup equity on 2003-04 vintages. While option ARMs have been the favored cashout product in 2006, that may change in 2007: First, qualification standards on option ARMs have tightened following the Final Interagency Guidance on Non-traditional Mortgages. Second, new hybrid neg-am mortgages offer better value to borrowers, and may become the preferred cashout product once it is more readily available. In purchase volumes, increased availability of fixed-rate IOs will likely take away share from hybrid IOs. Perhaps surprisingly, high combined loan-to-value (CLTV) and limited documentation loans may not decline significantly despite the slowdown in HPA at a national level, predominantly led by a slowdown in the hottest markets. In contrast, high CLTV loans have largely been originated in regions which experienced at or below average HPA, and originations in these products are likely to be relatively immune to the recent slowdown.
High CLTV products may not decline anytime soon
Figure 2. Non-agency Supply: Continued Shift Into Weaker Credit Loans Shows average collateral characteristics of prime supply.
Orig. Year 2004 2005 1Q 06 2Q 06 3Q 06 %IO 46 42 40 37 35 % Neg- % FICO %CLTV am <700 >90 13 28 33 35 40 36 37 44 39 41 15 15 16 17 20 %Limdoc 59 68 78 76 76 %Alt-B 16 17 17 18 21
Figure 3. Issuance Trends in Prime Structured MBS Shows composition of prime and sub-prime non-agency supply.
80 70 60 50 40 30 20 10 0 1Q04 3Q04 1Q05 Jumbo 3Q05 Alt-A Alt-B 1Q06 3Q06
Source: LoanPerformance, Lehman Brothers.
Source: Lehman Brothers.
December 12, 2006
31
The Surge in Demand for Floating Rate Assets Banks and CDOs dominate floater demand in prime MBS The relatively flat curve environment through 2006 created a strong demand for floating rate assets. As a result, over 50% of issuance in CMOs off fixed-rate collateral were floater deals (Figure 4). In agencies, strip and hard cap floaters were the norm, while in non-agencies, cap corridor floaters off alt-A collateral gained popularity. Traditional CMO investors banks and insurance companies were less active in 2006. Banks continued to pare down holdings of agency CMOs in favor of non-agency CMOs; Insurance companies shied away from long duration paper for most of the year while awaiting clarification on potentially severe provisions of FAS 155 regarding discount CMOs. Spread tightening in prime securities was largely driven by three players: banks with sub-LIBOR funding, Asian banks willing to express a view on short rates through capped floaters in agency CMOs, and most important, residential CDOs bidding on highyielding short duration assets. One such asset, AAA-rated cap corridor floaters off alt-A collateral offered at 50-60bp DM generated enough CDO demand to compress the 6% alt-A basis to only 10/32nd back of FN 6s in summer 06. Also notable is the role of CDOs in leading a convergence of credit spreads across sectors. For instance, BBBs off option ARMs, started the year at 300bp DM, significantly wider than sub-prime BBBs, and tightened rapidly to 150bp, only 10-15bp adrift of sub-prime. We expect CDOs to continue driving pricing in prime subordinates. Outlook for Demand in 2007 Inclusion of Hybrids in US Aggregate Index should benefit the sector While bank demand for securities has declined, we expect demand for whole-loans, especially in hybrids, to remain strong in a flat curve environment. The inclusion of hybrids in the U.S. Aggregate Index on April 1, 2007 is expected to lead to increased demand for both agency and non-agency hybrids from money managers. Post inclusion, hybrids will comprise 10% of the mortgage Index, and 3.5% of the U.S. Aggregate Index. Hybrids offer an attractive spread pickup versus 2-3yr agency debt, short duration corporates, and premium agency fixed-rate MBS.
Figure 4. Sharp Increase in Floater Issuance Shows floater deals as % of total agency and non-agency CMOs. Slope of 2s-10s shown on RHS in bp.
70% 60% 50% 40% 30% 20% 10% 0% 1Q01 1Q02 1Q03 1Q04 1Q05 1Q06 300 250 200 150 100 50 0 -50
Figure 5. Rising share of Prime Assets in CDOs
Composition of Assets backing Cash CDOs. 2004 Sub-prime MBS Prime MBS: AAA Prime Subordinates Total Prime MBS 57% 7% 14% 21% 2005 58% 8% 16% 24% 2006 56% 11% 17% 28%
Floaters
Slope 2s/10s (right, bp)

Source: LoanPerformance, Intex, Lehman Brothers.
December 12, 2006
32
CONVEXITY Dont Ignore Call Risk The flat curve significantly increases call risk With the last major refinancing wave in 2003-04 slipping into distant memory, the market appears to be disregarding call risk in mortgages. The issue is important given the call risk even in a $99 security in a flat curve environment. We discuss popular sources of call protection in our relative value section. We discuss the effect of HPA on prepayments, and recent surprises in seasoned hybrid prepayments below. Slower Housing Slows Discounts Prime turnover has slowed 34CPR in the hottest markets The effect of HPA on turnover is well known, and was evident in the recent housing slowdown of 2006. Indeed, prints on 50bp discounts slowed almost immediately. Figure 6 compares speeds at 12 wala on 2004 and 2005 vintages, with the latter experiencing significantly lower HPA over the first year. Speeds on 2005 vintage alt-A discount paper were slower by 3CPR, though they were less significant on jumbos. However, the impact was significantly greater in regions that experienced a larger slowdown in HPA. Figure 6 shows speeds in the top 25 percentile of MSAs, grouped by accumulated HPA over 2003-2005. These MSAs account for more than 60% of prime originations, and witnessed nearly a 20% slowdown in pace of HPA since 4Q05. The impact has been significant: speeds slowed by over 5CPR on alt-A discounts, and 3-4CPR on jumbos. And Also Premiums! Weak HPA significantly reduces callability, especially in weaker-credit pools The effect of HPA on refinancings however, is often overlooked. Figure 7 shows refinancing curves on conforming balance alt-A fixed-rates at 24 wala, by average annualized HPA on loans. Clearly, loans with lower built-up equity in a 5% HPA environment are 5-6CPR slower than in a 10% HPA environment. Moreover, the effect is even more pronounced for weaker credit loans in large part due to the higher share of high CLTV loans in high SATO pools.
Figure 6. Prime MBS Turnover Slowed Significantly in the Recent HPA Slowdown Shows speeds on 2004 and 2005 vintage, at 12 wala for 50bp discounts.
National Sector Agcy (FN) Alt-A 30yr Fxd 30yr Fxd 5/1 Hyb Jumbo 30yr Fxd 5/1 Hyb
th
Figure 7. Slower HPA Also Slows Refinancings Shows refinancing curves for low SATO (25-50bp) and high SATO (50-80bp) fixed-rates at 24 wala.
60 50 40
Top 25 Percentile Diff. -0.5 -3 -3 0 -2 04 14 16 23 13 17 05 12 12 18 10 14 Diff. -2 -5 -5 -3 -4
th
04 9.5 13 19 8 16
05 9 10 16 8 14
30 20 10 0 -100
-50
50
100
150
5% HPA, High SATO 5% HPA, Low SATO
10% HPA, High SATO 10% HPA, Low SATO
Top 25 percentile of MSAs selected based on accumulated HPA over 2003-2005. These hottest markets also saw the most significant slowdown in HPA over 4Q05 3Q06. Source: LoanPerformance, Agency pool data, Lehman Brothers.
December 12, 2006
33
Better Convexity of Fixed-rate IOs versus Level Pay Fixed rate IOs have more leveraged borrowers than LP The greater sensitivity of high CLTV loans to weaker HPA suggests that fixed-rate IO loans may be less negatively convex than level pay fixed-rates. With a larger purchase concentration and more leveraged loans, it is likely that fixed-rate IO borrowers may not be able to take advantage of refinancing opportunities as readily as level pay borrowers. Seasoned Hybrids: Recent Tail Prints May Be Misleading 2003 3/1s printed much faster than 2002 if adjusted for refi incentive The seasoned hybrid market has come a long way since last year, when hybrid valuations largely ignored value in the tail. However, recent evidence on tail speeds on agency 3/1s of 2002-03 have been encouraging: 3/1s of 2002 printed at 65-70CPR at reset and slowed notably in the months following reset. In early 2006, the 2003 vintage printed significantly slower: 55-60CPR on 4% 3/1s. This was largely due to in-the-money caps at first reset which restricted reset WACs to 6.5%, comparable to then prevailing mortgage rates. However, reset speeds on agency 3/1s in second half of 2006 have been much faster following the rally in mortgage rates. Indeed, speeds on 2003 vintage when adjusted for refinancing incentive have been significantly faster than 2002 vintage.
Figure 8. Lower Tail Speeds on Agency 3/1s of 2003 Shows hybrid speeds by months to reset (x-axis).
80 70 60 50 40 30 20 10 0 -18 -12 -6 0 6 12 18
Figure 9. Agency 3/1s of 2003 Printed Faster At Similar Refinancing Incentive Shows speeds at (-3, +3) months to reset, refinancing incentive measures versus 3/1 rate.
80 70 60 50 40 30 -150
-100
-50 2002 vintage
50 2003 vintage
100
2002 vintage
2003 vintage
Source: Agency pool data, Lehman Brothers
Source: Agency pool data, Lehman Brothers
December 12, 2006
34
PRIME CREDIT Uptick in Early Stage Delinquencies Performance of 2006 vintage Alt-B is 75% worse than 2005 Across products in the prime sector, the 2006 vintage has seen higher 90day+ delinquencies than the 2005 vintage, ranging from a 15-20% increase in alt-A fixed-rates, 25-30% in alt-A hybrids and a massive 75% increase in alt-B (Figure 10). There are three potential reasons: first, a shift in collateral composition towards weaker credits; second, the slowdown in housing; and third, deterioration in underwriting standards. We isolate the underwriting effect for alt-B collateral in Figure 11, by adjusting for slower HPA and collateral characteristics: we find that weaker underwriting accounts for over 60% of the underperformance on 2006 vintage alt-B collateral. HPA Effect on Delinquencies is Yet to Materialize Underwriting in Alt-B has clearly deteriorated Unfortunately, performance on 2006 vintage may get significantly worse. It takes about 8-9 months before slower HPA results in an observable increase in delinquencies; this is not yet fully captured in current performance trends. Thus, roughly 40% of underperformance of 2006 vintage alt-B collateral results from composition changes. Naturally, the question arises: How well do rating agency models capture the risk of collateral characteristics? We discuss specific properties below, but in general, the models underestimate risks in alt-B collateral. With rating agency subordination levels on alt-B that are close to prime/alt-A, and delinquency performance that is approaching levels seen only in sub-prime, we recommend moving up-in-credit on alt-B subordinates. Favor Alt-A and Option ARMs; Be Wary of Alt-B Subordination levels are aggressive on alt-B, conservative on option ARMs While rating agency levels are aggressive on alt-B, we find them fairly conservative on option ARMs. Of course, neg-am products are untested in a weak HPA environment and do merit conservative loss estimates. We discuss our approach in Appendix C, and find that, while it is difficult to make an outright call on option ARM subordinates, they are priced attractively versus sub-prime subordinates, and will likely outperform in a sustained housing downturn.
Figure 11. Performance adjusted for Deterioration in Borrower/Loan Characteristics and Slower HPA Shows 90day+ delinquencies and cumulative defaults on altB fixed-rates, adjusted for changes in collateral characteristics and HPA.
1.40 1.20 1.00 0.80 0.60 0.40 0.20 0.00 1 2 3 4 2004 5 2005 6 7 2006 8 9 1 2 3 4 2004 5 2005 6 7 2006 8 9
Figure 10. Alt-B Performance in 2006 is Significantly Worse Than in 2005 Shows 90day+ delinquencies and cumulative defaults on altB fixed-rates as % of original balance.
1.40 1.20 1.00 0.80 0.60 0.40 0.20 0.00
December 12, 2006
35
Subordination Levels: Aggressive on Layered Risks Subordination is aggressive on lower FICO, limited doc and high CLTV loans We compare rating agency subordination levels for different borrower/loan characteristics (Figure 12) with the historical sensitivity of defaults on fixed-rates to these characteristics across HPA scenarios (Figure 13). We conclude that: The risk in limited documentation loans (SISA, NINA 1) is understated, especially for lower FICO borrowers which forms a fairly large share of alt-B borrowers. Purchase loans require 20-25% lower subordination than cashout loans, even though performance has historically been worse due to a high share of first-time home buyers in purchase loans. Though rating agency subordination required for a first lien with a piggyback second has increased after model upgrades, it remains aggressive: and is only 25-40% higher than for standalone first liens. This is relevant given that high CLTV loans account for the majority of underperformance observed in 2006 vintage alt-B loans.
Subordination Levels by Borrower/Loan Characteristics
Figure 12.
Shows changes in subordination level for different characteristics for 5/1 hybrid IO. For instance, deals with 100% investor property loans would require 72% higher subordination at AAA level than deals with only owner-occupied loans, all else equal. Subordination Levels Borrower/Loan Characteristics Base Case
(1)
AAA 100 172 80 112 91 126
AA 100 170 78 111 91 124
A 100 175 76 112 91 125
BBB 100 179 75 112 91 125
Foreclosure Freq, % 0.86% 1.46% 0.65% 1.00% 0.68% 1.12%
Loss Severity, % 31% 36% 26% 31% 31% 31%
Investor property Purchase 680 FICO Full Documentation 80LTV w/ Piggyback

(1)
Base case corresponds to 720 FICO, owner occupied, cashout refinancing loan, no income, verified assets, 80LTV/CLTV. Source: Lehman Brothers, S&P Levels 5.7
Figure 13.
Cumulative Defaults on 2002 Vintage Alt-A Fixed-rates, %
Shows cumulative defaults for different characteristics, across HPA environments (annualized HPA). Low FICO is 650-675, high FICO is 725-750. 5% HPA Low FICO High FICO Low FICO High FICO Characteristics <80 CTV 80-100 CLTV <80 CTV 80-100 CLTV Full Doc Limited Doc Full Doc Limited Doc Non-Owner Owner Occupied Cashout Purchase
1
10% HPA 3.43 6.32 0.17 2.16 2.71 6.84 0.27 0.72 2.33 1.32 1.88 1.77
15% HPA 2.09 5.44 0.13 1.48 1.77 5.27 0.23 0.46 1.33 1.08 1.14 1.47
4.95 12.16 0.71 5.57 4.54 11.33 0.62 2.63 6.85 2.75 3.23 4.84
SISA: Stated Income, Stated Assets; NINA: No Income documentation, Not verified assets.
December 12, 2006
36
KEY TRADES IN PRIME MBS We summarize trade recommendations in Figure 14. Our key trades are:
1. Long Hybrids versus Fixed-rates
Supply demand technicals are likely to favor hybrid spreads in 2007: a decline in purchase volumes would hurt hybrid supply, while the inclusion of hybrids in MBS Index is likely to benefit the sector, as hybrids are good substitutes for 2-3yr agency debt, short duration corporates, and premium fixed-rate MBS.
2. Source Call Protection in Penalty Loans, Weaker Credit
The market is discounting call risk in mortgages, and we recommend sourcing call protection through penalty pools in hybrids, weaker credit conforming alt-A paper in fixed-rates, and premium California pools which are likely to see a significant slowdown in refinancings in a weaker HPA environment.
3. Sell Liquidity through Non-agency Basis
We recommend selling liquidity through non-agency basis, both in fixed-rates and hybrids. The fixed-rate jumbo basis at 25/32nd back of FN 6s picks nearly 15bp in OAS, as convexity of jumbos and TBAs narrowed when conforming loan sizes increased.
4. Credit: Favor Alt-A and Option ARM Subordinates
In prime credit, we favor moving up-in-credit: alt-A over alt-B, and option ARMs over sub-prime. In option ARMs, we recommend owning subordinates off XS/OC structures versus shifting interest structures, given the possibility of back-ended losses on this collateral. In alt-A shifting interest structures, we favor lower-rated tranches. In the following sections, we discuss call protection stories, evaluate floaters for cash and levered investors, and project performance of alt-A, alt-B and option ARM subordinates across housing scenarios.
Figure 14. Trade Recommendations in Prime MBS
Sector Floaters/ Cash Short Duration Trade Recommendation Favor AAA Mezz Option ARM floaters over HEL Own Agency 5/1 hybrids and 10yr CMBS vs. Agency TBA Non-agency 5/1 5.5 % vs. FN 6.5s Intermediate/Long Credit Own Non-agency 6.0s/6.5s vs. TBAs Theme: Down-in-Rating, Up-in-Borrower-Credit Own alt-A BBs vs. Subprime BBB-s hedged to be carry neutral (~1:2) Own Option ARM BBBs, Buy protection on subprime CDS Favor BBBs in XS/OC vs. shifting interest deals Based on Rating Agency Levels

Security Selection
Favor Cash-outs, Higher FICO Investor Properties Avoid 80/20s and Purchase Limited Doc loans (especially in Lower FICOs)
Call protection stories Are Under-priced: Penalties, Alt-A and California pools Seasoned hybrids with more than 15 months to reset still look attractive Favor Non-agency versus Agency hybrids Hybrids versus Short PACs/ Sequentials off Agency CMOs
December 12, 2006
37
#1. Call Protection Stories Favor penalty pools in hybrids, and conforming bal weaker credit pools in fixed-rates The call risk in mortgages is significant given the flat curve environment. Indeed, the 30bp rally in 10-year Treasury in November pushed the MBA refinancing index to its highest level in over a year, and prepayment speeds have picked up recently. We highlight the common call protection stories below: 3yr Penalty Pools: Premium penalty pools continue to trade at a substantial discount to fair value. $101 penalty pools in agency hybrids currently command a pay-up of only 6-7/32nd, while $102 penalty pools trade at a pay-up of 16/32nd. The pay-up can be monetized in premium pools through additional carry earned on penalty collateral. At 50bp in-the-money, premium penalty pools have a carry advantage of 1/32nd a month over non-penalty pools, at historical speed differentials of 15 CPR. At that rate, the current pay-up of 7/32nd for a $101 agency hybrid pool can be recouped in less than a year. Conforming Balance Alt-A: Although conforming balance loans command a 3-4/32nd pay-up over jumbos, the non-agency market typically does not price in any convexity benefit of weaker credit loans. Such conforming balance, weaker credit pools are expected to see muted refinancings especially in a slower housing market, and on a $101 pool, fair pay-ups are nearly twice as high as current pay-ups Premium California Pools: Over the past few years, California pools have seen faster turnover and higher callability than other geographies, on account of strong HPA. As the hottest HPA markets, including California, cool off rapidly, we expect to see a sharp decline in both turnover and refinancings, making 2006 vintage premium pools with 75%-80% California concentration very attractive. Seasoning: In $101 agency fixed-rates, the 2003 vintage typically trades at a 34/32nd pay-up over 2004 vintage on account of burnout. However, we favor 2004 production in non-agencies, as the 2003 vintage with 45%-50% accumulated HPA may see high cashout refinancings even in a scenario where housing is flat.
Figure 15. Source Call Protection through Weaker Credit, and Penalty Pools
Prepay differences at 50bp ITM Current Market Pay-up $100 3yr penalty Agency Hybrids 3yr penalty Non-agcy Hybrids 13-15CPR 10CPR 1-2/32nd $101 6-7/32nd $102 16/32nd $100 0-06 0-03 Pay-up at even OAS $101 0-15+ 0-06+ $102 1-06+ 0-08
Penalty, Non-agcy Fixed
5-8CPR
30% of Model Pay-ups
0-02
0-04+
0-05
Conforming Alt-A vs Jumbo Conf. Alt-A vs Jumbo Hybrids Premium Pools California: Fixed Premium Pools California: Hybrids
10CPR 8CPR 5-7CPR 5CPR
3-4/32nds pay-up 2-3bp tighter on N-spread None None
0-08 0-05
0-09 0-08 0-04+ 0-03
0-17 0-13 0-07+ 0-05
Prices as of 12/6/2006 are indicative only. Source: Lehman Brothers.
December 12, 2006
38
#2. Floating Rate Assets The flat curve environment has increased demand for floaters, which in turn has led to floaters created off a variety of CMOs structures and collateral. In addition to security lenders who monopolize the market for short floaters, there are three natural buyers who dominate the market for floaters with 3+ years of average life: CDOs: With floating rate liabilities, CDOs have shown a wide appetite for floating rate assets offering high yields, from AAA-rated cap corridor floaters to lower-rated subordinates. Bank demand for floaters has been primarily limited to regular strip floaters. With AA-rated banks funding at L15bp, floaters yielding Libor flat, or even L10bp are attractive short duration assets. Levered Vehicles such as hedge funds generate returns by increasing leverage to compensate for funding cost (20-50 times levered). Moreover, hedge funds often invest in AA and A floaters, earning additional spread for minimal credit risk.
Option ARM Floaters
Option ARM floaters offer 4-5 bp pick in OAS versus sub-prime floaters. While credit support in option ARMs is adequate, these floaters have been trading wide versus subprime largely as a result of headline risk. Within this asset class, cash players should favor AA and single-A floaters, off XS/OC structures. On an ROE basis, leveraged investors should favor AAA mezz MTA floaters levered 20-30x.
Hard Caps and Cap Corridors
Within agency CMOs, floaters with low strikes are trading rich and higher strikes look more attractive on an OAS basis. Alternatively, cash investors may favor non-agency hard cap floaters, with an addition 10bp in DM for selling liquidity. Cap corridor floaters off 6% alt-A fixed-rates offer the highest DM (60bp) and although they are structured with less AFC risk than six months ago, they still appear fairly rich on the model.
Figure 16. Comparison of Floaters
ROE (%) Sector Agency Cap Type Hard Coll. FN 6s Strike % WAL 7.0 7.5 FN 6.5s Non-Agency Hard Cap Corridor MTA Avail Funds 6.0s 6.0s MTA 7.5 7.0 8.5 9.0 9.0 9.0 9.0 Sub-prime Avail Funds HEL 9.0 9.0 9.0 3.0 3.0 3.0 3.0 3.0 3.0 3.0 5.0 5.0 3.0 5.0 5.0 Rating AAA AAA AAA AAA AAA AAA Mezz AA A AAA AA A DM 32 26 25 43 60 18 26 38 55 14 33 49 Option Cost LOAS (bp) Haircut (%) 36 27 24 35 55 1 1 0 0 2 2 2 -11 -5 -2 1 0 17 25 38 55 12 31 47 5 10 3-5 5 10 +5 +10 +3-5 +5 +10 3-5 +3-5 3-5 +3-5 2-3 Funding (L+ x) -2 No Vol 18.5 17.0 17.0 11.2 14.6 9.3 12.0 11.2 10.6 8.0 10.2 9.9 With Vol 0.5 3.5 5.0 4.2 4.0 9.0 11.7 11.2 10.6 7.3 9.8 9.6
Prices as of 12/1/06 are indicative only. Source: Lehman Brothers.
December 12, 2006
39
#3. Prime Subordinates In flat HPA, expect losses of 100bp on Alt-A, 200bp on AltB, and 205bp on option ARMs Most of the non-agency market, including leveraged products such as high CLTV IO products and option ARMs, have been largely untested in slow or negative HPA scenarios. Consequently, there is little historical data to anchor our loss estimates, and given this uncertainty, our estimates have a conservative bias. Figure 17 shows loss projections across collateral, for several constant HPA scenarios. In a flat HPA scenario, we project 104bp of losses on alt-A fixed-rate pool, and close to 200bp on both 2006 vintage alt-B, and 45bp SATO option ARM pool. 2 Note however that the alt-B sector is significantly more leveraged than option ARMs to negative HPA.
Move Up-In-Credit
For XS/OC structures in option ARMs, lower-rated tranches in shifting interest alt-A
We consider adequacy of BBB subordination across these HPA scenarios in Figure 18. While jumbo fixed-rates are well protected, alt-A fixed-rate BBBs take 10% loss in a -2% HPA scenario for life. At north of 200bp of losses, BBBs off Alt-B are completely wiped out in negative HPA scenarios. In option ARMs, deal structure plays an important role. Typically, XS/OC structures have OC floor of 50bp which protects investment grade subordinates against back-ended losses. In contrast, shifting interest structures often allow subordinates to pay down pro-rata after a 3-year period leaving BBBs exposed in scenarios where losses are back-ended.
Move Down-in-Rating in Shifting Interest Structures
Favor BB off alt-A fixed-rates versus BBB- subprime
The subordination mechanism in shifting interest structures thus suggests moving down in rating to pick spread while not significantly increasing principal losses. We favor BB vs BBB trade in alt-A fixed-rates/hybrids at current spread differentials of 350-400bp. Cross-sector, we recommend owning BBs off alt-A hedged carry neutral (1:2) with CDS on subprime BBB- and expect alt-A to outperform sub-prime in negative HPA scenarios.
Option ARMs: Cautious on Outright Exposure, Bullish on Relative Value
Favor BBBs off option ARMs over BBBs off subprime
While the steady deterioration in option ARM collateral characteristics makes us cautious on credit on an outright basis, we continue to favor BBB subordinates off XS/OC structures on a relative basis versus BBBs off sub-prime.
Figure 17. Projected Losses on Collateral Shows projected cumulative losses by collateral, for 2006 vintage. Projections on alt-B collateral are scaled 40% higher than implied by characteristics and HPA alone to account for weaker underwriting.
Lifetime HPA (%) Sector Jumbo Fixed Alt-A Fixed Alt-B Option ARM, 45bp
Figure 18. Comparison of Prime Subordinates Projected Tranche Losses on BBB subordinates. Shows performance of both shifting interest (SI) and XS/OC structures for option ARMs.
Lifetime HPA (%) +3% 0.21 0.72 1.32 1.38 Sector Jumbo Fixed Alt-A Fixed -4% 14 100 22 -2% 10 100 9 0% 7 64 3 +3% 4 23 0 -
-4% 0.52 1.74 3.56 3.16
-2% 0.42 1.46 2.94 2.61
0% 0.32 1.04 2.00 2.06
Alt-B Opt ARM, 45bp, SI Opt ARM, 45bp, XS/OC
Appendix C explains our approach to projecting losses on option ARMs.
December 12, 2006
40
APPENDIX A. SUPPLY/ DEMAND IN PRIME MBS
Sustained Cashout Activity Should Benefit Subprime and Option ARMs Shows share of cashout refi loans in various products. Nearly 5560% of loans in sub-prime and option ARMs are cashout refinancings. IOs are predominantly purchase products.
Prime Mortgages Orig. Year 2003 2004 2005 2006 Subprim e ARM 56 54 52 61 30yr Level Pay 26 27 36 33 5/1 Hybri d IO 25 20 26 27 Fixedrate IO 33 31 33 28 Option ARM 62 58 61 56
Gross Origination Volumes: The Strength of the Housing Market is As Important as Rates Agency fixed rate net issuance likely to pick up in a rally due to refinancings out of hybrids into fixed rates, especially fixed rate IOs.
Rate Scenarios (bp, Shift) Gross 12% HPA 5% HPA 0% HPA (100) 4,361 3,801 3,372 (50) 3,585 3,025 2,597 0 2,892 2,332 1,903 50 2,504 1,944 1,515 100 2,374 1,814 1,386
Fixed Net Issuance 12% HPA 5% HPA 0% HPA 370 298 225 323 246 169 289 207 125 267 180 92 253 160 66
Originations in Prime and IO/neg-am Products Have Been Concentrated in High HPA regions. Surprisingly, high CLTV originations are Not in Hottest Markets. Shows last 12 months (LTM) HPA at an MSA level weighted by gross origination volumes in prime and alt-A in each MSA.
25 20 15 10 5 0 4Q03
Demand: Banks Have Pared Down Agency CMO Holdings, and Favored Non-Agency CMOs. Dollar holdings of total securities by US Commercial and Savinfs banks shown on left axis. Agency and non-agency CMOs holdings as % of total securities portfolio shown on right axis.
$2.5 $2.0 $1.5 $1.0 $0.5 25% 20% 15% 10% 5% 0% 2006
2Q04
4Q04
2Q05
4Q05
2Q06
$0.0 2001
2002
2003
2004
2005
OFHEO LTM Neg-am
All Prime Orig CLTV>90
IO
Agency CMOs Total Securities

Source: FDIC.
Non-Agy CMOs
December 12, 2006
41
APPENDIX B: EFFECT OF SLOWER HPA ON PRIME PREPAYMENTS
Alt-A 5/1 Prepayments: Sensitivity to HPA Shows alt-A 5/1 hybrid prepayments at 24 wala, across annualized HPA scenarios (5%, 10%, 20% HPA).
90 80 70 60 50 40 30 20 10 0 -100 -50 0 5 10 50 20 100 150
Alt-B Fixed Rates: Sensitivity to Prepayments Shown at 24 wala, across HPA scenarios (5%, 10%, 20% HPA).
60 50 40 30 20 10 0 -100 -50 0 50 100 150
10
20
Option ARM Prepayments: 3yr Penalty Loans Have Been Printing Fairly Fast Shows prepayments by seasoning/wala on 3-year penalty option ARMs.
60 2002 45 2005 2004 2006
Largely Due to a Strong HPA Environment. Expect a 10CPR slowdown in prints for 10% slower HPA. Shows prepayments on 3-year penalty loans broken out by annualized HPA on loans.
60 4% HPA 20% HPA 12% HPA 24% HPA
45
30
30
15
15
0 0 6 12 18 24
0 0 6 12 18 24
Source: LoanPerformance.
Source: LoanPerformance.
December 12, 2006
42
APPENDIX C: PROJECTING LOSSES ON OPTION ARM COLLATERAL Given limited history of defaults on option ARMs, our estimates of potential losses for such products in different HPA environments is based on a two-step approach. First, we estimate losses on a pool of fixed-rate mortgages off a mix of collateral similar to that of the option ARM deal. Next, we scale up the estimate to account for potentially higher loss severities and a higher default frequency due to payment shocks on the option ARM product. We use SATO as a proxy for collateral credit quality we map collateral characteristics of an option ARM deal to fixed-rate products to calculate a fixed-rate equivalent SATO. We show the equivalent SATO for typical combinations of characteristics in 2006 option ARM deals below.
Credit Quality of Option ARM Collateral Has Deteriorated
Collateral Characteristics Mapped to Equivalent fixedrate SATO

% Lim Doc 91 92 87 95 87 90 94 92
FICO 2004 2005 1H 06 2H 06 5/1 IO Subprime 712 709 706 700 708 608
CLTV > % Non 80 Own 7 26 30 32 60 64 10 17 16 14 22 10
% Purchase 39 38 30 27 68 46
% Cashout 43 47 50 52 24 50
% Lim Doc 71 83 91 92 82 44
SATO 35 41 42 44 49 50 53 63
Margin % 2.75 3.50 3.25 3.50 3.25 3.25 3.50 3.25
FICO 734 702 705 706 695 703 704 715
Avg CLTV 73 74 78 78 78 79 78 84
% 2nd 28 0 37 36 29 41 31 63
% Purch 39 24 23 28 33 18 30 32
Greater Dispersion in Neg-am Pools With Potential for Recasts As Early As 18-20 mos Shows distribution of recasts by months from origination, assuming minimum payments, and rates move along forwards. Timing of Recast Vintage 2H 2005 Deal SATO, bp(!) 39 48 49 1H 2006 68 45 35
Option ARMs: Loss Predictions Across HPA based on Equivalent SATO. Typical Option ARM deals have 45bp SATO, and expected losses of 200bp in flat HPA scenario
Cum Loss Across HPA Scenarios, % SATO -4 2.09 3.16 4.03 -2 1.75 2.61 3.38 0 1.41 2.06 2.72 2 1.15 1.63 2.18 4 0.88 1.20 1.64
<24 mos 0% 0% 0% 39% 13% 13%
<30 mos 0% 43% 26% 81% 25% 36%
<36 mos 25 3% 85% 61% 100% 61% 54%

45 60
SATO on option ARM deal computed by estimating SATO of a fixed-rate mortgage with identical borrower/loan characteristics such as FICO, CLTV, occupancy, and purpose (purchase/cashout). Source: Lehman Brothers.
December 12, 2006
43
APPENDIX D. PRIME SUBORDINATES
Loss Projections on Prime Collateral by SATO Across HPA Scenarios
Projected Losses Across the Capital Structure in Alt-A: Even Higher Rated Subordinate Bonds Take Losses Tranche Loss %
Cum Loss % SATO Alt-A Hybrid Fixed Alt-B 25 30 100 1.59 1.63 2.81 1.16 1.19 2.10 0.95 0.97 1.73 0.75 0.76 1.38 -4 0 2 4 Alt-A Hyb AA A BBB BB 20 30 12 20 -4 -2
4 14
2 9
0 6
Projected Losses Across the Capital Structure in Alt-B: BBB Subordinates Take Losses; Favor Single A or Higher-rated Tranches Tranche Loss % -4 Alt-B AA AAA ABBB+ BBB0.0 0.0 0.0 33.4 100.0 100.0 0.0 0.0 0.0 0.0 74.5 100.0 0.0 0.0 0.0 0.0 14.9 100.0 0.0 0.0 0.0 0.0 0.0 100.0 0.0 0.0 0.0 0.0 0.0 97.5 -2 0 2 4
Projected Losses Across the Capital Structure in Option ARMs: Subordinates in XS/OC Structures Fare Better Than In Shifting Interest Structures
Tranche Principal Loss (%) Structure Shifting Interest Avg SATO 65 Rating AA A BBB XS/OC 65 AA A BBB Initial Sub (%) 5.10 3.80 2.80 3.95 1.60 1.10 -5 0 4 22 0 0 0 0 0 0 3 0 0 0 5 0 0 0 0 0 0
December 12, 2006
44
NON PRIME MBS
Navigating an Uncertain Credit Environment

2006 RecapCredit Concerns Regained Focus Credit concerns regained focus in 2006 Credit concerns took center stage in the subprime sector over the past year as weaker underwriting on the 2005/06 vintages started to show up in early-stage performance and concerns about weak housing fundamentals got amplified through the year. This triggered an upsurge in short interest from macro and cross-sector hedge funds that overwhelmed buying interest from a record CDO new issue pipeline ($180 billion), leading subordinate spreads wider over the second half of the year. Strong two-way flows helped the synthetic market maintain its exponential growth trajectory, as about $150-$200 billion of BBB/BBB- notional traded in the single-name CDS and ABX market. As we had expected, the maturity in synthetics along with weaker credit performance resulted in an increase in tiering, although it remained confined largely to the synthetics market. Key Themes for 2007Negative Headlines to Dominate and are expected to get amplified in 2007 We expect credit issues from 2006 to be further amplified in 2007, which should result in significant negative headlines in the sector on multiple dimensionsweaker credit and prepayment performance on 2005/06 originations because of worse underwriting and a slower housing market, significant uptick in downgrade activity on 2005/06 issuance, and pressure on originator/servicer profitability because of lower volumes and worse credit. We advocate an overall defensive posture in 2007, and recommend moving up the capital structure. However, we continue to expect tiering to increase in 2007 based on crossvintage, capital structure, and credit views, which should provide significant opportunities for intra-sector relative value. We highlight our key trade recommendations below.
HEL Portfolio Positioning in 2007

Views Basis Capital Structure Trade Overweight 1-3yr AAA Floaters versus Consumer ABS Move to AAA/AAs from single-As/BBBs, outright short BBB-s Long HEL BBBs, short CDO single-As/AAs carry neutral Favor 2005 versus 2006 and 2004, 2H03 and 1H04 look the weakest Favor cashout heavy deals vs. purchase, overweight full doc heavy deals Long high IO% deals from the 2004-1H05 vintage Synthetics Rationale 5-20 bp pickup in carry, minimal AFC risk (1-2bp), supply technicals look positive, GSE sponsorship to continue. Potential weakening in CDO demand following headline data around prepays/credit and downgrades. Outright short on 2H06 BBB- attractive based on current spreads implying about a 15% probability of negative HPA. Fair spread on CDO single-A/AAs much wider than current levels because of high leverage to HPA. Significantly lower HPA leverage as a result of built-up equity on 2005, weak tails on 2H03/1H04 because of negative excess spread, high FRM% and passing triggers, 2006 underwriting looks weak. Rating agency subordination levels for purchase loans and limited documentation loans are lower than historical experience. Rating agencies do not factor in faster post-reset speeds on IOs by 10%-15% CPR as a result of higher FICOs, loan balance, payment shocks and built-up equity.
Cross Sector
Cross Vintage
Credit Stories
Box trade: Long 06-2 BBB/06-1 BBB- paired BBB/BBB- basis on ABX 06-2 is tighter than fair value, neutralizing the HPA with short 06-2 BBB-/06-1 BBB directionality by pairing with an opposite view on 06-1. Long BB+, short BBB- carry neutral as out of the money protection to weak housing Good alternative to shorting single-A outright, trade breaks even in cashflow terms below 0HPA, does not rely on mark to market P&L vs. single-A short
December 12, 2006
45
KEY THEMES FOR 2007 Will Credit Assumptions Be Revised Higher by CDO Buyers? potentially leading to increased risk aversion from CDO buyers The million dollar question for 2007 is whether end-CDO liability buyers will step back from the market given increased credit concerns. Following prior widening episodes, we have seen CDO liability demand remaining fairly strong, which has driven HEL subordinate spreads tighter because of attractive equity returns at wider asset spread levels. We believe that there are two drivers that might potentially change this in 2007 and increase risk aversion among CDO liability buyers. The market should see confirmation of the expected weakening in performance in a weak HPA scenario in observed headline performance on the 2005/06 vintages. In a 0% HPA scenario, we expect prepayments on the 2006 vintage to be about 15%20% CPR slower than 2003 experience at similar seasoning (Figure 1a). In addition, 2006 vintage delinquencies should see about a 100% increase versus 2003 (Figure 1b). Slower observed prepayments and higher delinquencies should revise loss assumptions higher. We expect a sharp increase in the number of bonds at the risk of downgrade over the next year (Figure 2). We base these estimates off weak early credit performance for the 2005/06 vintage, increasing aggressiveness of the rating agencies in downgrading 2005/06 vintage bonds early based on credit performance, and weak loss coverage multiple profiles for the 2003/04 vintage tails. Thus, while actual bond losses are likely back-ended post-2007, an increase in downgrades might cause adverse structural (related to OC triggers) and mark-to-market issues for CDOs.
Figure 1a. ARM Prepay Projections in 1H07, 0% HPA

1Q07 Vintage 1H05 2H05 1H06 2H06 Proj. 69% 35% 29% 14% 2003 73% 46% 51% 28% Chg. -4% -11% -22% -14% Proj. 63% 41% 28% 17% 2Q07 2003 67% 55% 48% 41% Chg. -4% -14% -20% -24% Vintage 1H05 2H05 1H06 2H06 Proj. 17.3% 12.9% 10.8% 5.2%
Figure 1b. ARM 60+ Dq Projections in 1H07, 0% HPA

1Q07 2003 13.5% 8.7% 5.4% 2.6% Chg. 3.8% 4.2% 5.4% 2.6% Proj. 23.0% 15.1% 13.5% 8.9% 2Q07 2003 19.1% 10.2% 7.2% 4.3% Chg. 3.9% 4.9% 6.3% 4.7%
Source: Lehman Brothers, Loan Performance
Figure 2.
1000 800 600 400 200 0
Historical Downgrades, Projected Number of Bonds at Risk of D/G
1H03
2H03
1H04
2H04
1H05
2H05
1H06
2H06
1H07
2H07
Actual Dow ngrades

Source: Lehman Brothers, Moodys.
5% HPA
0% HPA
December 12, 2006
46
Continued Pressure on Originators/ServicersWACs to Rise Further and keeping originator/servicer profitability under pressure We expect originator profitability to remain under pressure over the next year for two reasons (Figure 3a). First, supply is expected to be lower by 15%-20% versus 2006 levels, driven by a slower housing market, tightening in underwriting standards, and increase in WACs as a result of a repricing of credit risk premium higher (we estimate about a 30-50 bp increase in WACs). Second, although gain-on-sale margins might remain depressed because of credit issues and excess capacity in the origination channel, further cost reduction potential is limited. Thus, given the expected increase in headline news around originator viability, we would argue for tiering between originators based on the financial strength of the sponsor as an additional theme in 2007. In addition to originators, the servicer community is also expected to see a large increase in the dollar volume of delinquent loans to be serviced over next year. We expect about a 70% increase in the balance of 60+ delinquent loans over the next 12 months, which will likely strain the current servicing capacity in the system (Figure 3b). Also, as 60+ delinquent loans are expected to increase to about 11% of the outstanding universe, increase in servicing costs is expected to pressure financial viability and the ability to maintain servicing quality relating to loss mitigation.
Figure 3a. Gain on Sale Margins, by Quarter

5% 4% 3% 2% 1% 0% 1Q02
Figure 3b. Projected Delinquent Loans, by Date

180 150 120 90 60 30 12% 10% 8% 6% 4% 2% 0% Jan-03 Industry 60+ Dq, $bn
1Q03 Prime
1Q04 Alt A
1Q05
1Q06
0 Jan-01
Jan-05
Jan-07
Subprime
% Outstanding (Right)
Performance Projections for 2005/06 Originations 2005/06 performance expected to be 40% worse than 2003 because of worse underwriting While the deterioration of early stage performance for 2005/06 originations versus prior vintages has been well documented, there are two broad questions relating to 2005/06 performance that will become clearer over the next year. How will the 2005/06 vintage eventually perform, i.e., is the 2005/06 vintage underperformance largely an EPD issue, or does tail credit look poor as well? Based on 2005 vintage roll rates (current to 30-59 day) being consistently higher than the 2003 vintage by 40%-50% even at higher WALAs (12-21 months), we conclude that the underwriting of the remaining pool looks poor as well, and the 2005/06 underperformance is not purely an EPD issue. Assuming that these roll rate differences persist, we estimate that lifetime cumulative defaults on 2005/06 to be about 40% higher than 2003 (for similar characteristics and HPA), with both tail 60+ delinquencies (% cur. bal.) and CDRs higher by 3%-5%. What are the largest drivers of 2005-06 underperformanceis risk layering to blame? We find that a large part of the early-stage underperformance compared with prior vintages can be attributed to the high CLTV segment (Figures 4a and 4b). Thus, product types with the highest proportion of high CLTVssecond liens,
47
December 12, 2006
purchase, and hybrid loanshave underperformed relatively more than cashout and fixed-rate loans. To examine the effect of risk layering, we find that within the high CLTV universe, performance2005 vintage versus 2003 vintagehas worsened similarly for both purchase and cashout borrowers, as well as for full documentation and low documentation loans. Thus, high-CLTV as a characteristic in itself has been a good proxy for stretched borrowers, and additional risk layering on the purchase and low documentation dimensions has not had a significant additive effect. However, within the high CLTV universe, the relative underperformance of the low SATO borrowers2005 versus 2003has been about twice the high SATO population. This is also supported by origination trends, as the increase in share of low SATO loans in high CLTV originations has increased to about 50% in 2005 from 37% in 2003, possibly indicating that lenders have been aggressive in expanding in the high FICO, alt-B segment potentially at the expense of underwriting.
Figure 4a. <70 CLTV, Adjusted 60+ Dq by Vintage
5.0% 4.0% 3.0% 2.0% 1.0% 0.0% 3 6 2002 2003 9 2004 12 2005 15
Figure 4b. >90 CLTV, Adjusted 60+ Dq by Vintage

8.0% 6.0% 4.0% 2.0% 0.0% 3 6 2002 9 2003 2004 12 2005 15
Source: Lehman Brothers, Loan Performance. Shows 60+ delinquencies adjusted for FICO, CLTV, HPA, documentation, purpose, loan type, loan size, SATO, unemployment, foreclosure timelines across vintages.
Large Balance of Hybrid Resets Ahead Significant balance of resets next year, but not the largest performance issue for 2007 We project about half of the current outstanding subprime hybrids ($270 billion of the $600 billion outstanding in the securitized universe) to come up to their first reset date over the next year (Figure 5a). Based on the observed performance on the $200 billion of loans that have reset higher over the past 18-24 months, we expect the following for postreset performance in 2007. Given the large incentive of 2004/05 vintage hybrids borrowers to refinance postreset, and the ability to do so because of accumulated equity, we expect an additional 6% CPR increase in post-reset prepayments as a result of payment shocks. However, the combination of adverse selection and payment shocks should increase hybrid default rates to 22%-24% CDR and 60+ delinquencies to about 35% in the tail (Figure 5b). Post-reset performance for high CLTV and low HPA borrowers has been poorer historically because of fewer available options to bailout delinquent borrowers. The additional effect of payment shocks on post-reset prepayments is only about 3%-4% CPR, but delinquencies can rise to about 40%. This might be a significant issue for the 2H05 hybrids in late-2007 under a flat/declining housing scenario.
Thus, while adverse selection and higher defaults post-reset will start to weigh on tail 2003/04 vintage deal performance in 2007, we do not expect payment shocks to be the largest performance issue in the subprime market over the next year.
December 12, 2006 48
Lehman Brothers | U.S. Securitized Products Research Figure 5a. Projected Balance ($bn) of Hybrid Resets
40 30 20 10 16% 12% 8% 4%
Figure 5b. 60+ Dq. Across Annualized Shock, by WALA

40% 30% 20% 10% 0% 3 6 9 12 15 18 21 24 27 30 33 36 No Shock 9% Shock (Annualized)
0 0% Jan-98 Jan-00 Jan-02 Jan-04 Jan-06 Jan-08 Jan-10 ARM Reset IO Reset Annual Shock , %(Right)
New Synthetic Products to Increase Levered Money Involvement Expect further maturity in synthetics with emergence of new correlation products We continue to expect further growth and maturity of the synthetic market in 2007, with the two major themes being the emergence of correlation benchmarks through ABX tranches and increase in new levered vehicles such as constant proportion debt obligations (CPDOs) and derivative product companies (DPCs). Similar to the corporate CDS market, we expect the development of a tranche market on the ABX indices to be the next stage of product development in the ABS synthetics. The largest impact of an active ABX tranche market would be to provide a pricing and correlation benchmark for managed CDO liabilities and bespoke tranches. Dealers are currently working to roll out a standardized set of documents by late-January 2007, which will likely use the two most recent ABX indices as the underlying portfolio. Hedge funds might find the product attractive for cross-capital structure correlation trades because of more transparent pricing of the underlying, absence of senior management fees and lack of OC/IC trigger specification in the tranche cash flows. However, the presence of a natural long investor base for ABX tranches other than correlation desks is questionable given the lack of a rating, which will exclude CDOs as a buyer base. In addition to correlation products, we expect to see the replication of corporate CPDO technology to the ABS markets. The CPDO vehicle is structured to sell protection on an underlying index using significant leverage and is able to issue AAA rated liabilities at attractive L+100-200 returns. The issuance of these trances has had a significant impact on tightening spreads in the corporate market over the past few months as a result of the high leverage. Dealers are trying to replicate a similar structure on an underlying portfolio of BBB/BBB- ABX indices using about 2x to 5x times leverage, which should add to the investor base of the ABX indices and might potentially provide a backstop to spreads.
December 12, 2006
49
PORTFOLIO POSITIONING IN 2007 Basis Call: Overweight 1-3 year AAA Floaters versus Consumer ABS 1-3yr AAAs look attractive vs. consumer ABS because of carry pickup and positive technicals We believe that 1-3 year AAA HEL floaters look attractive versus the consumer ABS sectors as a result of a 5-15 bp pickup in carry with minimal additional risk. Technicals look positive for next year as supply should be lower by 15-20% and demand is likely to shift to short cash flows up the capital structure in a housing slowdown. From a fundamental standpoint, AFC risk is minimal (1-3 bp) and bonds are well protected from downgrades even under a housing meltdown scenario (-5% HPA over life). Cross Capital Structure: Move to AAA/AAs from Single-A/BBBs Move up the credit spectrum to defend against headline risk We advocate a defensive posture by moving up the capital structure because of negative headline news around weaker credit/prepayment performance, a sharp uptick in downgrade activity, and pressure on originator/servicer profitability. This might potentially dampen the CDO bid for assets, which should favor the AAA/AA stack versus single-A/BBBs (where the marginal bid is largely from CDOs). Additionally, based on current BBB- spread levels in the ABX/CDS market (350bp), we believe that the market is pricing in about a 15% probability of negative home price appreciation, with a mean HPA of about 3%. Given our base case view of 0% HPA for 2007, we continue to believe that being outright short at the BBB- level looks attractive. Cross Vintage: Overweight 2005 versus 2004 and 2006 2005 subordinates attractive versus 04 and 06 vintages We find that 1H05/2H05 subordinates look attractive versus both the 2004 and 2006 vintages after factoring in differences in underwriting, built-up equity and structure (Figure 6). Although 2003/04 deals have lower HPA leverage, the tail of outstanding subordinates look weak because of high fixed-rate concentrations, low excess spread, and weak triggers that lead to release of over collateralization (OC) despite negative excess spread in the tail. The 2006 vintage has the highest leverage to HPA, and cumulative defaults are expected to be about 40% higher than 2003 because of weaker underwriting. The 2005 vintage underwriting while still worse than 2003, is about 10% better than 2006 and has about 12-24 months of built up equity, which significantly reduces HPA leverage.
Figure 6. Loss Coverage Multiples across Vintages, 2% HPA
LCM A2 A3 Baa1 Baa2 Baa3 1H03 2.12 1.59 1.36 1.19 1.03 2H03 2.46 1.93 1.40 1.03 0.83 1H04 1.88 1.41 1.12 0.88 0.75 2H04 3.08 1.65 1.31 1.10 0.96 1H05 6.06 3.94 2.54 1.30 1.18 2H05 4.34 3.15 2.18 1.55 1.23 1H06 2.79 2.16 1.67 1.26 1.00 2H06 2.54 1.93 1.53 1.21 0.98
Source: Lehman Brothers. Loss coverage multiple is defined as the multiple of the base case default curve at which the bond takes a first dollar of principal writedown.
Credit Stories: Historical Performance versus Subordination Differences Purchase and limited doc loans look under enhanced Although most collateral-based differences are reflected in rating agency credit models, we find a few areas of disconnect between historical credit performance and loss coverage requirements. Purchase loans and limited documentation loans look under enhanced: We find that purchase loans have historically underperformed cashout after adjusting for other characteristics. However, rating agencies require lower subordination for purchase loans relative to cashout (Figure 7). Also, while limited documentation loans have
50
December 12, 2006
about 20% higher loss coverage requirements, we find this to be inadequate given about 35% higher cumulative defaults historically for limited documentation loans relative to full documentation. Following the recent changes from the rating agencies regarding becoming more conservative on high CLTV loans, we find that the additional subordination for high CLTV loans seems fair versus historical performance from the 2000-2004 vintages. However, given the recent underperformance in the high CLTV universe in the 2005/06 vintages and expected underperformance in a weak HPA environment, we remain vary of deals with high 100 CLTV concentrations.
Figure 7. Rating Agency Foreclosure Frequency versus Historical Cum Defaults

Rating Agency Foreclosure Frequency 100* 137 100 117 100 154 Historical Cum Defaults @ 24 WALA Multiples 100* 78 100 136 100 153 Cum Defaults 3.6% 2.8% 3.6% 4.9% 3.6% 5.5%
Loan Characteristics Purchase Cashout Full Doc Lim Doc 80 LTV 80 LTV/100 CLTV
Source: Lehman Brothers, Standard and Poors LEVELs 5.7 * Foreclosure frequency for a purchase 80 LTV,80 CLTV, full doc loan is benchmarked to 100
Synthetics: Play the BBB/BBB- Basis Using a Box Trade Expect the BBB/BBB- basis on ABX 06-2 to widen more than ABX 06-1 Based on running the underlying portfolio of the ABX 06-1 and 06-2 indices across HPA scenarios, we back out the implied HPA distributions at each rating level given current market spreads (Figure 8). For example, the ABX 06-2 BBB- spread of 370 bp implies that the market is pricing in a mean HPA of 3.15% HPA with the probability of less than 0% HPA being 14.7%. Thus, we recommend being long the index with the lowest market implied HPA (market is pricing in a conservative HPA scenario) and shorting the index with the highest implied HPA. Based on the results in Figure 7, we recommend being short the ABX 06-2 BBB- versus the 06-2 BBB. However, this trade is a negative credit directional trade and is exposed to overall HPA coming in stronger than expected. Thus, to neutralize any directionality in the trade to HPA and express a pure view on mispricing across the capital structure, we would recommend pairing it with an opposite view on the 06-1 index, i.e., short 06-1 BBB and long 06-1 BBB-. Thus, the view is that in an overall widening the BBB/BBB- basis on the 06-2 would widen more than the basis on the 06-1, and in a tightening the 06-2 basis would tighten less than the 06-1.
Figure 8. Implied HPA at Current Market Spreads
ABX Series ABX HE 06-1 ABX HE 06-1 ABX HE 06-2 ABX HE 06-2 Current Rating Spread BBB BBBBBB BBB145 270 235 370 Market Implied Mean HPA 1.52% 1.74 2.42 3.15 Market Implied Probability of <0 HPA 30.7% 28.1 21.0 14.7 Market Implied Tranche Loss (%) 3.3% 6.5 7.5 11.0
December 12, 2006
51
ABS CDO OUTLOOK
Focus on Fundamentals
2006 RECAP Supply The ABS CDO market experienced yet another year of record-breaking issuance The ABS CDO market experienced yet another year of record-breaking issuance. Through November, total ABS CDO issuance was $177 billion, compared with $80 billion in 2005. Unlike previous years in which high grade CDO issuance dominated the market (71% in 2005), 2006 experienced a dramatic increase in mezzanine issuanceto 45% of overall supply (Figure 1a). This was due in large part to the easy access hybrid/synthetic mezzanine transactions have to sourcing assets via the singlename CDS markets. Additionally, as the high grade arbitrage became less attractive throughout the year (due to tightening of asset spreads), high grade supply tapered off relative to mezzanine (Figure 1b). The mezzanine arbitrage, on the other hand, remained relatively strong. Spreads In terms of spread action, after a short-lived spread widening event in January, the high grade and mezzanine markets entered into a one-directional trade through the summer months with spreads tightening across the capital structure (Figure 2). The voracious search for yield brought banks, money managers, insurance companies, and other CDOs into the sector, easily absorbing the record supply. However, a combination of a year-end supply deluge and continued softness in housing fundamentals began pushing spreads wider heading in 4Q. Spreads widened from their August tights by 2 bp for AAAs (25-32 bp), 3 bp for AAs (52-54 bp), 0-14 bp for As (134-149 bp), and 24-38 bp for BBBs (312-344).3
Figure 1a ABS CDO Issuance

Issuance ($bn) 18 16 14 12 10 8 6 4 2 Jan % 80% 70% 60% 50% 40% 30% 20% 10% 0% Mar May Jul Sept Nov
Figure 1b Theoretical Equity IRR

IRR (%) 25% 20% 15% 10% 5% 0% Jan Feb Mar Apr May Jun Jul Aug Sept Oct Nov High Grade Mezzanine
Mezzanine
High Grade
% High Grade (RHS)
Source: Lehman Brothers. Assumes zero loss and no change in collateral composition throughout the year.
As of 11/30/06.
December 12, 2006
52
Lehman Brothers | U.S. Securitized Products Research Figure 2 ABS CDO Spreads
High Grade AAA 12/31/05 Current Change Max Min 27 25 -2 27 23 AA 57 52 -5 59 49 A 149 134 -15 162 135 BBB 291 312 21 332 288 12/31/05 Current Change Max Min AAA 31 32 1 36 30 Mezzanine AA 60 54 -6 67 51 A 163 149 -14 171 135 BBB 342 344 2 385 306
Source: Lehman Brothers. Current spreads as of 11/30/06.
2007 OUTLOOK AND KEY THEMES Focus on the Fundamentals Much of our recent research has focused on home price appreciation sensitivities on HEL securities and, ultimately, ABS CDOs (see High Grade and Mezzanine: A Housing Story). This analysis is useful to understand the range of HPA scenarios that may result in CDO collateral losses. However, this has been more of an extrapolation of historical data because there is little data in a 0% or lower HPA environment. When do we expect our HPA scenarios to become reality? Therefore, an important question for 2007 is when do we expect these scenarios to become reality? Investors will need to keep a vigilant eye on changes in the underlying fundamentals of the mortgage loans in order to track whether performance is better or worse than expectations. We expect worse underwriting trends in 2005 and 2006 vintages, as well as slower HPA to promulgate itself through slower voluntary prepayments and higher delinquencies in those vintages. With this in mind, we summarize the performance trends we expect 2005 and 2006 vintage HEL pools to experience by mid-2007 under a 0% HPA assumption, focusing on voluntary prepayments and 60-plus delinquencies (we also provide direct links to our ABS Analyzer so investors can track performance of these vintages easily going forward): Voluntary prepayments (CRR): The slowdown in HPA is expected to result in significantly slower prepayments on 2006 vintage HELs, the vintage with the least amount of built-up HPA. The 2005 vintage should also experience slower prepayments, but not to the same extent. For instance, the 2006 prepayments are expected to be in the 15-25 CPR range by 2Q07. This is about a 20 CPR decline versus the 2003 vintage experience. By 2Q07, 1H05 prepayments are expected to be about 60 CPR and 2H05 prepayments are expected to be 40 CPR. This is lower than the 2003 vintage by roughly 3 CPR and 10 CPR, respectively. Delinquencies: Slower HPA is also expected to have a significant impact on delinquencies, with the 2006 vintage ARMs approaching 6%-11% by 2Q07, almost doubling the 2003 experience. The 1H05 and 2H05 vintage ARMs will likely trend up to 18% and 13% by 2Q07, respectively. This is a 30%-50% increase versus 2003. We anticipate CDO liabilities repricing the additional risk of CDO collateral losses The decrease in voluntary prepayments and increase in delinquencies would result in higher cumulative loss expectations on HEL pools going forward, compared with recent experience. As a result, we would anticipate that this would lead to CDO liabilities repricing the additional risk of CDO collateral losses.
December 12, 2006
53
Tiering to Increase Recent 2005 and 2006 vintage HELs are not necessarily in the clear for 2007 from a rating downgrade perspective At the risk of sounding like a broken record, we believe tiering will increase in 2007. This should be familiar to clients who read last years outlook and our report Gearing for Tiering. As it turned out, 2006 started out with greater tiering within the BBB space (measured by spread variance), but quickly diminished as the CDO bid for CDO liabilities gained steam. Looking forward, however, we expect tiering to increase as a result of a number of factors. Growth in CDS on CDOs: Using the HEL market as a guide, increased activity in the single-name CDS market should lead to increased tiering among CDO transactions. Investors will have the ability to express both positive and negative views on specific transactions. Increased Sophistication in Valuation Techniques: As the ABS CDO market continues to develop new valuation techniques, differences between CDO transactions will become more apparent. Currently, uncovering these differences can be a tedious exercise to say the least. Deal-specific Performance: Historically speaking, rating agencies were not active in downgrading or upgrading HEL securities until after the three-year stepdown date. Given the age of the 2003 and 2004 vintages, downgrades within these vintages will increase in 2007, especially if HPA slows further. This could have a more immediate effect on seasoned CDOs that have exposure to those vintages (Figure 3). For example, the 2004 and 2005 vintage mezzanine ABS CDOs have 36% and 23% exposure to the 2004 HEL vintage, respectively. However, more recent 2005 and 2006 vintage HELs are not necessarily in the clear for 2007 because select downgrade actions within these vintages have already occurred in 4Q06.
Figure 3. Mezzanine ABS CDO HEL Vintage Exposure

HEL Vintage Exposure CDO Vintage 2003 2004 2005 2006
2003 24% 14 3 0
2004 8% 36 23 7
2005 7% 3 32 50
2006 5% 3 1 19
Total 45% 56 60 76
Growth in CDS of CDOs In 2006, a standardized template for CDS on CDOs (see Standardized PAUG CDS on CDOs) was introduced, creating a more active market for CDS on CDOs. We expect the market to continue to grow in 2007, fueled by CDO managers desire to source CDO assets more selectively (new-issue and seasoned) and interest from investors looking to buy or sell protection on subordinate tranches. We expect investors to express more actively the capital arbitrage trades and paired trades that take advantage of structural and collateral differences. A more standardized ABX tranched market would quickly become a benchmark from which ABS CDO liabilities are compared An interesting development will come from two new markets: ABX tranches and bespoke ABS CDOs. The ABX tranches are likely to become more actively traded beginning in 1Q07. A more standardized tranched market will quickly become a benchmark from which ABS CDO tranches are compared, just as the ABX indices have become a more liquid benchmark for single-name CDS. This is not to say the comparison of synthetic ABX tranches to managed CDOs will be an easy one, but it is likely to result in greater convergence. Additionally, a more active be-spoke market will add another entre to the menu. These additional tools will make understanding structure and
54
December 12, 2006
We advocate moving up in credit in both high grade and mezzanine transactions
collateral ever more important in determining relative value and will also put the spotlight on CDO managers ability to differentiate themselves as to how they create value for investors. From a hedging perspective, the development of a more actively traded tranched and be-spoke market will provide CDO investors with additional options for hedging positions. Relative Value Up in credit: We advocate moving up in credit in both high grade and mezzanine transactions. Given the uncertainty regarding the outlook for HPA over the coming year, we do not believe current pricing fully incorporates the downside risk from any deviation from the base case. Focus on structure: While our overall bias is to be up in credit, we do believe there are significant opportunities within a given rating based solely on structural differences. We recommend taking a closer look at differences between traditional trigger, trigger holiday, and triggerless structures. Long/short trades would make sense in this regard.
December 12, 2006
55
Explanation of the Lehman Brothers Mortgage Model The Lehman Brothers Mortgage Valuation Model allows investors to analyze mortgage-backed (MBS), asset-backed (ABS) and commercial mortgage-backed securities (CMBS). The model collects pertinent and material information needed to evaluate and calculate the risk measures of the security. The model provides option-adjusted spreads and durations along with other risk measures using Lehman Brothers' Prepayment, Default, and Term Structure Models. Analyst Certification The views expressed in this report accurately reflect the personal views of Srinivas Modukuri, Prasanth Subramanian, Neil Barve, Jose John, Michael Lee and Brian Zola, the primary analysts responsible for this report, about the subject securities or issuers referred to herein, and no part of such analysts' compensation was, is or will be directly or indirectly related to the specific recommendations or views expressed herein. Important Disclosures Lehman Brothers Inc. 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. Lehman Brothers generally does and seeks to do investment banking and other business with the companies discussed in its research reports. As a result, investors should be aware that the firm may have a conflict of interest. To the extent that any historical pricing information was obtained from Lehman Brothers 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. Lehman Brothers' global policy for managing conflicts of interest in connection with investment research is available at www.lehman.com/researchconflictspolicy. To obtain copies of fixed income research reports published by Lehman Brothers please contact Valerie Monchi (vmonchi@lehman.com; 212-526-3173) or clients may go to https://live.lehman.com/. Company-Specific Disclosures Lehman Brothers Inc. or one of its affiliates has managed or co-managed a public offering of securities or Rule 144A offering of securities for the following issuers within the past twelve months: the Federal Home Loan Mortgage Corporation (FHLMC), the Federal National Mortgage Association (FNMA), and the Government National Mortgage Association (GNMA). These companies are current investment banking clients of Lehman Brothers or companies for which Lehman Brothers would like to perform investment banking services. Legal Disclaimer This material has been prepared and/or issued by Lehman Brothers Inc., member SIPC, and/or one of its affiliates ("Lehman Brothers"). Lehman Brothers Inc. accepts responsibility for the content of this material in connection with its distribution in the United States. This material has been approved by Lehman Brothers International (Europe), authorised and regulated by the Financial Services Authority, in connection with its distribution in the European Economic Area. This material is distributed in Japan by Lehman Brothers Japan Inc., and in Hong Kong by Lehman Brothers Asia Limited. This material is distributed in Australia by Lehman Brothers Australia Pty Limited, and in Singapore by Lehman Brothers Inc., Singapore Branch ("LBIS"). Where this material is distributed by LBIS, please note that it is intended for general circulation only and the recommendations contained herein do not take into account the specific investment objectives, financial situation or particular needs of any particular person. An investor should consult his Lehman Brothers' representative regarding the suitability of the product and take into account his specific investment objectives, financial situation or particular needs before he makes a commitment to purchase the investment product. This material is distributed in Korea by Lehman Brothers International (Europe) Seoul Branch. Any U.S. person who receives this material and places an order as result of information contained herein should do so only through Lehman Brothers Inc. This document is for information purposes only and it should not be regarded as an offer to sell or as a solicitation of an offer to buy the securities or other instruments mentioned in it. With exception of the disclosures relating to Lehman Brothers, this report is based on current public information that Lehman Brothers considers reliable, but we do not represent that this information, including any third party information, is accurate or complete and it should not be relied upon as such. It is provided with the understanding that Lehman Brothers is not acting in a fiduciary capacity. Opinions expressed herein reflect the opinion of Lehman Brothers' Fixed Income Research Department and are subject to change without notice. The products mentioned in this document may not be eligible for sale in some states or countries, and they may not be suitable for all types of investors. If an investor has any doubts about product suitability, he should consult his Lehman Brothers representative. The value of and the income produced by products may fluctuate, so that an investor may get back less than he invested. Value and income may be adversely affected by exchange rates, interest rates, or other factors. Past performance is not necessarily indicative of future results. If a product is income producing, part of the capital invested may be used to pay that income. Lehman Brothers may, from time to time, perform investment banking or other services for, or solicit investment banking or other business from any company mentioned in this document. No part of this document may be reproduced in any manner without the written permission of Lehman Brothers. 2006 Lehman Brothers. All rights reserved. Additional information is available on request. Please contact a Lehman Brothers' entity in your home jurisdiction.

(Lehman Brothers) Mortgage Outlook For 2007 - Bracing For A Credit Downturn

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

(Lehman Brothers) Mortgage Outlook For 2007 - Bracing For A Credit Downturn

Uploaded by

Copyright:

Available Formats

Fixed Income Research U.S.

Securitized Products Research

December 12, 2006

Mortgage Outlook for 2007: Bracing for a Credit Downturn

Lehman Brothers | U.S. Securitized Products Research

December 12, 2006

Lehman Brothers | U.S. Securitized Products Research

December 12, 2006

Lehman Brothers | U.S. Securitized Products Research

CROSS SECTOR SUMMARY RECOMMENDATIONS

December 12, 2006

Lehman Brothers | U.S. Securitized Products Research

Figure 1 Attribution of MBS Index Returns

Excess Returns 72 (34) 121 (23) 67

Carry + Convexity Losses 70 (59) 38 0 2

(10) (12) (77) (11) (12) (14)

49 25 (140) (27) (33) (35)

8 (72) (438) (84) (92) (84)

FHT 231 - 2005 5.5% FNT 361 - 2005 6.0%

Performance Attributed to Curve (bp)

December 12, 2006

Lehman Brothers | U.S. Securitized Products Research

New borrowers in the system has grown

Figure 3 Number of Loans in Fixed Rates Increasing

Figure 4 Paydowns are Returning with Larger Loansizes

Source: Lehman Brothers

Source: Lehman Brothers

December 12, 2006

Lehman Brothers | U.S. Securitized Products Research

Source: Attribution by Lehman Brothers. Total Net Issuance from EMBS

-50 323 246 169

0 289 207 125

100 253 160 66

370 298 225

December 12, 2006

Lehman Brothers | U.S. Securitized Products Research

Overseas Demand Should Stay Strong

Central bank asset reallocation should continue

Private overseas institutions have stepped away from treasuries as well

December 12, 2006

Lehman Brothers | U.S. Securitized Products Research

Figure 8. Share of Assets of Foreign Central Banks

Source: US Treasury, Lehman Brothers

Treasury Investments of Private Overseas Investors Have Stalled

Source: Federal Reserve Flow of Funds

Growth Rates of Bank Assets

December 12, 2006

Lehman Brothers | U.S. Securitized Products Research

Margin from Holding Mortgages Is Low & Likely to Decline

Historical Spread Betw een Mortgage Y ields and Cost of Funds

Corporate Spreads vs. Mortgages

Curve A djusted ZV of 100bp Discount vs. 5yr A Corp Sprd to Tsy

Lehman Brothers | U.S. Securitized Products Research

Convexity Profile of the MSR Univ ($B 10yrs)

Prof ile of MSR as of Dec 8th

Prof ile of MSR as of July 14

December 12, 2006

Lehman Brothers | U.S. Securitized Products Research

Call protection is being discounted

Seasoning Curves of 50bp Discounts (1m % CPR)

December 12, 2006

Lehman Brothers | U.S. Securitized Products Research