United States

Mortgage Strategies

Alt-A M a r k e t: An I n tr o d u c ti o n
As originators look for ways to increase market share, they are finding the Alt-A market an attractive choice. Many different characteristics classify loans as Alt-A, and they each have an impact on the performance of Alt-A securities. The definition of Alt-A loans has evolved over time from high credit subprime/low credit prime to include characteristics such as non-owner occupancy, less than full documentation, and others. The Alt-A market has experienced tremendous growth, increasing over 300% in the past two years. Alt-A MBS have better convexity than prime and agency MBS: Specifically, in-the-money loans are less likely to prepay owing to impaired credit of some kind, whereas out-of-the money loans are more likely to prepay because of credit curing and defaults. Investor properties have historically prepaid more slowly, and have had lower delinquencies and cumulative defaults than owner-occupied properties.

Jennifer Cohen Jennifer.Cohen@gs.com New York: 212-357-2280 Limin Zhang Limin.Zhang@gs.com New York: 212-902-9281

Goldman Sachs

When considering prepayment and default curves for Alt-A collateral, adjustments are made to the base case in congruence with different characteristics. For prepayments, the largest shift comes from loan size and occupancy. For defaults, it comes from FICO, LTV and documentation.

While documentation level alone appears not to have a direct impact on performance, this characteristic is almost always layered with other riskmitigating factors such as low balance, high FICO, and low LTV, which coupled together have better convexity.

Mortgage Strategies

United States

Overview
The Alt-A market emerged in 1996, as investors started to better understand the credit characteristics of mortgagors and as originators began expanding the spectrum of credit they would lend to. Prime originators expanded their score cards to accommodate lower FICO scores, while subprime originators expanded theirs to accommodate higher ones. As the market has evolved over time, different outlying characteristics began to define the Alt-A universe, such as documentation levels, occupancy, and other risklayering attributes. Today, the guidelines for what constitutes an Alt-A loan are constantly changing and vary by issuer. Some originators view Alt-A loans as falling into one of two categories: Prime Alt-A, which consist of mortgages to borrowers who are prime but who are missing documentation (those who are self-employed, have no employer references, etc.), or to borrowers who will not occupy the subject property; or Non-Prime Alt-A, which consist of mortgages to strong subprime borrowers.

Other participants in the market divide the categories even further. FitchRatings defines traditional Alt-A loans as those with agency-conforming insured balances and a prime credit history, but with expanded guidelines for other categories including loan purpose, debt-to-income ratios, and loan-to-value ratios. FitchRatings also divides the Alt-A universe into Prime Alt-A, A-AltA, and Alt-B loans in an attempt to group together perceived and real distinctions in the credit risk of pools 1 within each category. Exhibit 1 summarizes the criteria for these three buckets. Exhibit 1: FitchRatings Alt-A categories Category FICO LTV % Investor Property Stated/No Documentation Typical Issuers Prime Alt-A 700 730 60% - 70% 10% - 30% 30% - 50% RALI, BOA, Countrywide CB and TI, Wells Fargo A-Alt-A 670 700 65% - 85% < 30% 50 75% Impac, CSFB Alt-B 640 670 < 80% 5% - 30% < 75% First Franklin, Chase, Ameriquest, New Century, Meritage Second Liens, Prepayment Penalties

Other
Source: FitchRatings, Mortgage Principal and Interest, May 2004.

FitchRatings, Mortgage Principal and Interest, Tug of War: New Dimensions of the Alternative A Market. May 2004.

Goldman Sachs

United States

Mortgage Strategies

Countrywide has defined Alt-A in a different way. Deals issued off of the CWALT T1 shelf include jumbo loans with low documentation or jumbo loans with full documentation but lower-than-prime FICO scores (<680). Deals issued off of the CWALT T2 shelf are composed of jumbo low and no documentation loans. The CWALT CB shelf houses collateral that is closer to prime in credit quality, but has conforming balances. The other category in Exhibit 2 below consists of mostly J series securities, which are Countrywides conduit deals. Exhibit 2: Different Countrywide shelf characteristics (2003-3Q2004)
T2 $1,502m 683 FICO 75 LTV 85% Low/No Docs $458K Avg. Balance

Other (Mostly "J" Series) $11,111m 703 FICO 74 LTV 52% No/Low Docs $231K Avg. Balance

T1 $4,004m 687 FICO 68 LTV 56% Low/No Docs $490K Avg. Balance

CB $14,452m 718 FICO 75 LTV 47% Low/No Docs $166K Avg. Balance

Sources: Loan Performance, Goldman, Sachs & Co. Data as of October 2004. Please note that all amounts are based on Loan Performance available deals.

As a final example, the Impac Alt-A program has a different mix of characteristics that it classifies as Alt-A. In general, Impacs Alt-A loans have these characteristics: low FICO scores, high LTVs, and no or low documentation, low loan balances, underlying properties that are usually owner-occupied, and prepayment penalties.

coupled with:

Clearly, different issuers have different Alt-A definitions. Exhibit 3 contains a summary of different general characteristics for a number of AltA issuers. The information was taken from the Loan Performance database, which is estimated to capture 85% of the non-agency Alt-A market.

Goldman Sachs

Mortgage Strategies

United States

Exhibit 3: Alt-A characteristics by issuer

Issuer Accredit HL Adj Rate Mtg Am Hm Mtg Ameriquest BA Mortgage Bear Stearns Mtg Sec Capstead Chase Citigroup Countrywide Fund/Hm CSFB Deutsche DLJ DSLA First Franklin GE Capital Mortgage GSAA Harborview Headlands Homestar ICIFC IMPAC Indy Mac/RAST Lehman MALT Morgan Stanley Norwest Integrated PNC PNCMT RFC RALI RFC RAMP SAMII SASCO Soundview WA Mutl Wells ALT A # of Deals 1 2 2 1 27 42 1 5 2 96 65 9 1 1 3 2 2 7 11 5 3 53 127 1 41 8 6 28 1 150 19 3 103 1 8 3 Original # of Loans 3,692 4,624 6,402 2,943 72,733 87,710 982 8,427 8,974 215,552 106,993 15,593 662 2,762 13,493 2,987 8,570 16,531 14,973 8,398 5,132 126,078 180,840 1,433 88,853 20,457 9,753 78,872 5,436 283,045 52,077 1,407 190,028 1,543 13,473 3,352 ARM/ Fixed ARM ARM ARM Fixed Fixed ARM/ Fixed Fixed Fixed Fixed Fixed ARM/ Fixed Fixed Fixed ARM Fixed ARM ARM/ Fixed ARM Fixed ARM/ Fixed Fixed ARM/ Fixed Fixed Fixed Fixed ARM/ Fixed Fixed Fixed Fixed Fixed Fixed Fixed ARM/ Fixed ARM/ Fixed ARM Fixed FICO 713 714 706 723 731 702 N/A 723 703 706 692 695 N/A 703 648 N/A N/A 718 712 694 N/A 693 707 717 711 724 725 697 N/A 719 705 707 716 672 715 717 LTV 76 76 78 67 69 77 78 74 76 74 77 79 73 72 83 75 N/A 74 75 77 79 79 72 72 74 72 77 77 81 75 102 91 72 73 74 74 Occupancy (Majority) Owner Owner Owner Owner Investor Owner Owner/ Investor Owner Owner Owner Owner Owner Investor Owner Owner Owner Owner/ Investor Owner Owner Owner Owner Owner Owner Owner Owner/ Investor Owner Owner/ Investor Owner Owner Owner/ Investor Owner Owner/ nvestor Owner Owner Owner Owner/ Investor Doc Low Low Full/Low/ None Full Low Low Low Full Low Low Low Low Low Low Full Full/Low Full/Low Low Low Low N/A Low Low Full/Low/ None Full/Low Low Full Low N/A Full/Low Full Full/Low Full/Low Full/Low Low Full/Low

Sources: Loan Performance, Goldman, Sachs & Co. Data as of October 2004.

Goldman Sachs

United States

Mortgage Strategies

Over time, the largest shift in Alt-A characteristics has been in loan size, percentage of ARM loans, and percentage of loans with a prepayment penalty. In 1996, only 1% of loans originated were ARMs; however, in 2004, almost 70% of loans originated were ARMs. It is interesting to note that both prime jumbo and subprime loans had the same directional change. Exhibit 4: Shift in non-agency MBS characteristics by outstandings 1996
Original Volume ($mn) WA FICO WA LTV ARM % % Investor Property % Limited or No Documentation WAC Average Loan Size % Prepayment Penalty % Primary Mortgage Insurance Alt-A 5,496 706 74 1% 22% 51% 8.88 133,858 0% 13% Prime 21,550 713 76 12% 0% 19% 7.75 283,505 1% 17% Subprime 10,512 646 78 36% 5% 18% 9.28 66,440 13% 6% Alt-A 26,254 709 75 69% 20% 61% 5.62 225,334 21% 14%

2004
Prime 56,481 735 68 73% 3% 42% 4.70 425,787 8% 2% Subprime 44,326 632 82 72% 5% 35% 7.06 149,248 49% 19%

Sources: Loan Performance, Goldman, Sachs & Co.

Alt-A originations have seen a general shift over time in occupancy, FICO scores, and documentation. Exhibit 5 shows that credit scores have stayed relatively stagnant in deals defined as Alt-A, dipping slightly in 2000 but lately improving; meanwhile, occupancy and documentation levels have shifted. It is interesting to note that in 2000-2002, originating Alt-A loans tended to have higher owner occupancy levels and lower documentation levels.

Goldman Sachs

Mortgage Strategies

United States

Exhibit 5: FICO, occupancy, and documentation changes, 1997-2004

Alt-A FICO 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 1997 1998 1999 2000 2001 2002 2003 2004 820+ 740 - 820 660 - 740 <660

Alt-A Documentation
100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 1997 1998 1999 2000 Full 2001 2002 2003 2004 Low None 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 1997 1998

Alt-A Occupancy

1999

2000

2001

2002

2003

2004

Owner Occupied

Non-Owner/ Investor

Second Home

Sources: Loan Performance, Goldman, Sachs & Co. Data as of June 2004.

Goldman Sachs

United States

Mortgage Strategies

Issuers and underwriters in the Alt-A market are constantly changing as well, partly owing to the ongoing entrance of new players in the market. Exhibit 6 and Exhibit 7 show the Alt-A issuers and underwriters with the highest dollar volume of Alt-A securities issued during the first three quarters of 2004. Exhibit 6: Top Alt-A issuers in 2003 and 2004 (Jan-Sept); $, millions Rank
1 2 3 4 5 6 7 8 9 10

Issuer
Countrywide Financial Impac Bear Stearns CS First Boston/ABSC IndyMac UBS Warburg Lehman Brothers GMAC-RFC American Home Mortgage Chevy Chase Bank Top 10 Sub-Total Market Total

2004
$17,561.6 14,753.7 10,752.7 10,484.0 10,342.2 9,515.6 6,234.8 4,580.3 4,332.3 4,076.7 92,633.9 $110,988.7

Volume 2003
$10,758.9 4,857.7 4,730.2 6,061.0 4,380.7 3,233.6 5,740.0 6,981.9 0.0 2,251.6 48,995.6 $55,953.0

Change
63.2% 203.7 127.3 73.0 136.1 194.3 8.6 -34.4 NA 81.1 89.1 98.4%

2004 Deals Market Share

30 11 18 9 19 20 19 17 4 4 151 208

15.8% 13.3 9.7 9.4 9.3 8.6 5.6 4.1 3.9 3.7 83.5 100.0%

Source: Inside MBS & ABS, November 12, 2004

Exhibit 7: Top Alt-A underwriters in 2003 and 2004 (Jan-Sept); $, millions Rank
1 2 3 4 5 6 7 8 9 10

Issuer Bear Stearns RBS Greenwich Capital Markets Countrywide Securities Corp. CS First Boston UBS Warburg Lehman Brothers Banc of America Securities Goldman Sachs Deutsche Bank Citigroup / Solomon

2004
$22,587.9 15,585.2 15,405.0 14,813.6 11,651.6 9,783.3 3,796.5 3,602.0 3,111.8 2,851.4 103,188.3 $110,988.7

Volume 2003
$11,673.6 4,295.0 10,590.3 8,605.5 5,386.7 6,227.5 4,156.2 35.5 1,132.7 1,842.4 53,945.4 $55,953.0

Change
93.5% 262.9 45.5 72.1 116.3 57.1 -8.7 NA 174.7 54.8 91.3 98.4%

2004 Deals Market Share

48 28

20.4% 14.0 13.9 13.3 10.5 8.8 3.4

29 17 29 25 13 20 11 17 2372 208

Top 10 Sub-Total Market Total

3.2 2.8 2.6 93.0 100.0%

Source: Inside MBS & ABS, November 12, 2004

Includes co-mandates, hence some deals are double counted.

7

Goldman Sachs

Mortgage Strategies

United States

Why the Alt-A Market is attractive Growth

The dollar volume of issuance in the Alt-A market has increased dramatically, rising from $10.95 billion at the end of 1Q2002 to $47.96 billion at the end of 3Q2004 more than a 300% increase. Exhibit 8: Alt-A issuance 1Q2002-3Q2004 ($, billions)
60 50 40 30 20 10 1Q02 2Q02 3Q02 4Q02 1Q03 2Q03 3Q03 4Q03 1Q04 2Q04 3Q04

Source: Inside MBS & ABS, November 12, 2004.

This explosive growth is expected to continue. In order to maintain volume levels as refinancing opportunities decline, prime originators will expand downward in the credit spectrum while subprime originators will expand upward, meeting at the AltA market. Alt-A originations amounted to an estimated $90 billion for the first half of 2004, an increase of 136.8% from the first half of 2003. As the Alt-A market expands, top subprime originators are currently developing AltA programs and underwriting guidelines; however, to currently manage their programs, they have begun to carve out higher FICO originations and sell them separately from their typical originations. Potential true Alt-A entrants include Ameriquest, New Century, and Fremont. Compared with prime outstandings, the number of Alt-A loans has continued to grow while prime originations have had dips in production.

Goldman Sachs

United States

Mortgage Strategies

Exhibit 9: Alt-A, jumbo prime, and subprime outstandings, 19972004 as estimated by Loan Performance ($, millions)
400,000 350,000 300,000 250,000 200,000 150,000 100,000 50,000 0 1997 1998 1999 2000 2001 2002 2003 2004

Alt-A
Sources: Loan Performance, Goldman, Sachs & Co.

Subprime

Jumbo

Convexity
When they are in the money, Alt-A securities typically prepay more slowly than nonagency prime securities. For prime Alt-A securities, prepayments are usually slow in the first year and mimic prime prepayments in later years. For non-prime Alt-A securities, prepayments are typically similar to those of subprime initially, or slightly faster (since they do have better credit quality); then they speed up if their credit quality has improved and the borrower can refinance at a prime rate. Although default levels for Alt-A securities are higher than those of jumbo MBS, they are not nearly as high as default levels for MBS with subprime collateral. So, Alt-A securities have higher coupons to compensate for this risk, though it has proven to be minimal. As a result, the Alt-A market enjoys lower prepayment risk than a prime pool, and lower default risk than a subprime pool. The convexity of Alt-A securities is less negative than that of agency MBS. Specifically, when rates are low, fewer Alt-A borrowers refinance (because they have credit, document, and other problems) this is also known as call protection. When rates are high, prepayments are higher for Alt-A securities than for agency MBS owing to defaults and credit curing this is also called extension protection. A sophisticated investor base Now that Alt-A characteristics are becoming better understood, investors have gained a clearer idea of the risks they are taking, which means they have become more willing to pay for the higher coupon the Alt-A collateral produces.
3

It is estimated that Loan Performance captures 85% of outstanding non-agency MBS.

9

Goldman Sachs

Mortgage Strategies

United States

Historical performance of Alt-A securities

In practice, different risk factors are layered for Alt-A securities, so certain concerns are mitigated. But in this initial primer on the subject, we examine each different AltA characteristic on its own.

Prepayments
The behavior of prepayment curves for Alt-A securities by vintage is similar to what we would expect for prepayment curves of other mortgage products. To examine historical Alt-A prepayments, we first need to determine current coupons for fixed loans of various vintages. It was determined that the weighted average current coupon for fixed, 30-year Alt-A loans was 8.29% in 1999, 9.24% in 2000, 8.05%in 2001, and 7.37% in 2002. Using these weight averages, we determine current coupon buckets by vintage and gather prepayment data along these lines. Exhibit 10 shows the results of this exercise. Historically, the vast majority of Alt-A loans have been fixed, so we will examine only this subset in our analysis. Exhibit 10: Alt-A prepayments by vintage
1999 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% Jan-99
90% 80% 70% 60% 50% 40% 30% 20% 10% 0% Jan-00 2000

3 Month CPR

Jan-00

Jan-01

Jan-02

Jan-03

Jan-04

3 Month CPR

Jan-01

Jan-02

Jan-03

Jan-04

< 7.75% Bucket > 8.75% Bucket Subprime Current Coupon 2001 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% Jan-01

7.75 - 8.75% Bucket Jumbo Prime Current Coupon

< 8.75% Bucket > 9.75% Bucket Subprime Current Coupon 2002 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% Jan-02

8.75 - 9.75% Bucket Jumbo Prime Current Coupon

3 Month CPR

Jan-02

Jan-03

Jan-04

3 Month CPR

Jan-03

Jan-04 6.75 - 7.75% Bucket Jumbo Prime Current Coupon

< 7.5% Bucket > 8.5% Bucket Subprime Current Coupon

7.5 - 8.5% Bucket Jumbo Prime Current Coupon

< 6.75% Bucket > 7.75% Bucket Subprime Current Coupon

Sources: Loan Performance, Goldman, Sachs & Co.

10

Goldman Sachs

United States

Mortgage Strategies

Based on these different vintage and bucket combinations, we can examine the characteristics of these pools. As Exhibit 11 shows, FICO scores of Alt-A borrowers have increased slightly over time for current coupon and below-market coupon borrowers. The lower boundaries of FICO scores, which indicate lower borrower quality, have stayed the same. In each vintage/WAC bucket scenario, the lower WAC buckets have higher FICO scores, which would be expected since better borrowers receive lower rates. LTVs have the same pattern, and also seem to have stayed relatively stagnant over time. Exhibit 11: Alt-A characteristics by vintage and WAC bucket
Original Investor Balance 2,644,896,981 5,915,758,953 3,115,151,593 3,895,561,394 5,657,511,655 2,527,643,773 7,584,342,572 10,285,046,960 5,911,968,438 10,116,213,119 14,819,499,105 8,771,051,840 Original Number of Loans 15,393 39,612 21,466 16,369 30,852 14,745 25,975 41,732 30,423 36,171 64,798 48,067 Avg Loan Size 171,825 149,343 145,120 237,984 183,376 171,424 291,986 246,455 194,326 279,677 228,703 182,476 Original WA Credit Score 705 697 686 712 699 681 719 702 679 725 711 683 Original WA Combined LTV 74 77 82 76 81 86 74 79 87 71 78 87

1999

< 7.75% Bucket 7.75 - 8.75% Bucket > 8.75% Bucket

2000

< 8.75% Bucket 8.75 - 9.75% Bucket > 9.75% Bucket

2001

< 7.5% Bucket 7.5 - 8.5% Bucket > 8.5% Bucket

2002

< 6.75% Bucket 6.75 - 7.75% Bucket > 7.75% Bucket

Sources: Loan Performance, Goldman, Sachs & Co.

Goldman Sachs

11

Mortgage Strategies

United States

Exhibit 12 highlights Alt-A prepayment performance by loan size. As would be expected, larger loans are more sensitive to changes in the interest rate environment: When rates drop, borrowers with high mortgage payments have an incentive to refinance leading to higher prepayments. When rates rise, it becomes more expensive to re-locate or buy a more expensive house, so people tend not to move and prepayments decline. Exhibit 12: Alt-A loan size prepayment speeds by vintage and WAC
1999 Vintage, 7.75 - 8.75% WAC Bucket 80% 70% 60% 3 Month CPR 50% 40% 30% 20% 10% 0% Jan-99 Jan-00 Jan-01 Jan-02 Jan-03 Jan-04
3 Month CPR 80% 70% 60% 50% 40% 30% 20% 10% 0% Jan-00 Jan-01 Jan-02 Jan-03 Jan-04 1M+ 2000 Vintage, 8.75 - 9.75% WAC Bucket

0+ - 300k

300k+ - 600k

600k+ - 1M

0+ - 300k

300k+ - 600k

600k+ - 1M

2001 Vintage, 7.50 - 8.50% WAC Bucket 80% 70% 60% 3 Month CPR 50% 40% 30% 20% 10% 0% Jan-01 0+ - 300k Jan-02 Jan-03 Jan-04 600k+ - 1M 1M+
3 Month CPR 80% 70% 60% 50% 40% 30% 20% 10% 0% Jan-02

2002 Vintage, 6.75 - 7.75% WAC Bucket

Jan-03 300k+ - 600k

Jan-04 600k+ - 1M 1M+

300k+ - 600k

0+ - 300k

Sources: Loan Performance, Goldman, Sachs & Co.

12

Goldman Sachs

United States

Mortgage Strategies

FICO score is an indicator of credit. Exhibit 13 shows Alt-A prepayment performance by FICO score. In falling interest rate environments, higher FICO score loans have historically prepaid slightly faster. This could possibly reflect the greater sophistication of the borrowers and the options for refinancing that are most available. Exhibit 13: Alt-A FICO score prepayment speeds by vintage and WAC
1999 Vintage, 7.75 - 8.75% WAC Bucket 70% 60% 3 Month CPR 3 Month CPR 50% 40% 30% 20% 10% 0% Jan-99 Jan-00 600 - 659 Jan-01 Jan-02 660 - 719 Jan-03 Jan-04 70% 60% 50% 40% 30% 20% 10% 0% Jan-00 Jan-01 600 - 659 Jan-02 Jan-03 Jan-04 720 - 779 2000 Vintage, 8.75 - 9.75% WAC Bucket

720 - 779

660 - 719

2001 Vintage, 7.50 - 8.50% WAC Bucket 70% 60% 3 Month CPR 3 Month CPR 50% 40% 30% 20% 10% 0% Jan-01 Jan-02 600 - 659 Jan-03 660 - 719 Jan-04 720 - 779 70% 60% 50% 40% 30% 20% 10% 0% Jan-02

2002 Vintage, 6.75 - 7.75% WAC Bucket

Jan-03 600 - 659 660 - 719

Jan-04 720 - 779

Sources: Loan Performance, Goldman, Sachs & Co.

Goldman Sachs

13

Mortgage Strategies

United States

Exhibit 14 demonstrates Alt-A prepayment performance by LTV. As would be expected, higher LTV loans have historically prepaid more slowly, as fewer refinancing options are available to the borrowers. Also, the borrowers have less equity in their homes, so they are less likely to move. Exhibit 14: Alt-A LTV Prepayment speeds by vintage and WAC
1999 Vintage, 7.75 - 8.75% WAC Bucket 80% 70% 60% 3 Month CPR 50% 40% 30% 20% 10% 0% Jan-99 <60 Jan-00 Jan-01 Jan-02 Jan-03 80+ - 90 Jan-04 > 90 3 Month CPR 80% 70% 60% 50% 40% 30% 20% 10% 0% Jan-00 <60 Jan-01 60+ - 70 Jan-02 70+ - 80 Jan-03 Jan-04 > 90 2000 Vintage, 8.75 - 9.75% WAC Bucket

60+ - 70

70+ - 80

80+ - 90

2001 Vintage, 7.50 - 8.50% WAC Bucket 80% 70% 3 Month CPR

2002 Vintage, 6.75 - 7.75% WAC Bucket 80% 70% 60% 3 Month CPR 50% 40% 30% 20% 10%

60% 50% 40% 30% 20% 10% 0% Jan-01 <60 Jan-02 60+ - 70 Jan-03 70+ - 80 Jan-04 80+ - 90 > 90

0% Jan-02 <60

Jan-03 60+ - 70 70+ - 80

Jan-04 80+ - 90 > 90

Sources: Loan Performance, Goldman, Sachs & Co.

14

Goldman Sachs

United States

Mortgage Strategies

As mentioned in the introduction of this report, one typical characteristic of Alt-A borrowers is that they are taking out a mortgage on a property that they will use as an investment or a second home (such properties are commonly called non-owner occupied). Deals made up of this type of collateral can offer investors a prime type borrower with a higher coupon (to adjust for the non-owner occupancy) and slower prepayments (investors are less likely to engage in the hassle of refinancing and they will not relocate). As Exhibit 15 shows, the vintage WAC buckets with the higher WACs have a higher percentage of non-owner occupied properties. Exhibit 15: Alt-A occupancy type by vintage and WAC bucket Owner Occupied 86.5% 75.2% 72.1% 90.6% 79.9% 82.6% 94.1% 89.0% 85.2% 85.7% 83.0% 81.8% Second Home 2.7% 2.8% 2.6% 2.1% 2.5% 2.2% 1.7% 2.3% 2.4% 2.0% 2.2% 2.5% Non-Owner Occupied 10.8% 22.0% 25.3% 7.3% 2.5% 15.2% 4.2% 8.7% 12.4% 12.3% 14.8% 15.7%

1999

< 7.75% Bucket 7.75 - 8.75% Bucket > 8.75% Bucket

2000

< 8.75% Bucket 8.75 - 9.75% Bucket > 9.75% Bucket

2001

< 7.5% Bucket 7.5 - 8.5% Bucket > 8.5% Bucket

2002

< 6.75% Bucket 6.75 - 7.75% Bucket > 7.75% Bucket

Sources: Loan Performance, Goldman, Sachs & Co.

Even when the incentive is there, non-owner occupied properties are less likely to experience prepayment, partly owing to the inconvenience and time that refinancing involves. Also, investors in such properties do not have many alternatives for obtaining financing. In addition, the relocation component of this investor base is very small, as the owners are not living on the subject property (or live there only part time).

Goldman Sachs

15

Mortgage Strategies

United States

Exhibit 16 looks at 12-month CPR (June 2003 June 2004) for deals that have a higher percentage of properties that are used as investments (or investor owned properties, part of the non-owner occupied universe). Note that if we look at the investor owned universe and the total Alt-A universe in the loan performance database, the investor owned properties deals have historically prepaid more slowly Exhibit 16: Sample of Alt-A deals where investor owned properties make up more than 33% of total % Investor Owned
74.16% 73.81% 64.15% 63.52% 63.51% 60.38% 58.63% 43.72% 38.19% 37.31% 36.26% 33.09% 53.84% 16.07% 16.77%

Deal
BAMALT 2003-1 BAMALT 2003-4 BAMALT 2003-2 BAMALT 2003-5 BOA Funding-2 BAMALT 2003-3 BAMALT 2003-6 Master ALT 2003-01 Master ALT 2003-02 Countrywide2002AL8 RFC 2002-QS11 Countrywide2002AL15 Total < 1/3% Investor Owned Alt-A Total Alt-A

12 Month CPR
29% 17% 27% 14% 38% 21% 12% 42% 41% 49% 44% 47% 31% 47% 47%

Original Balance
326,022,467 569,523,193 605,393,831 459,054,279 150,266,740 301,176,362 504,754,939 328,533,613 580,836,253 502,262,670 240,263,928 460,582,846 5,028,671,121 266,139,498,601 271,168,169,722

WAC
6.49 5.95 6.37 5.89 6.64 6.25 5.92 7.20 6.92 7.41 7.27 7.66 6.63 7.08 7.07

Age
20 15 18 14 21 16 13 21 19 27 24 23 19 21 21

Sources: Loan Performance, Goldman, Sachs & Co.

16

Goldman Sachs

United States

Mortgage Strategies

Looking at the current coupon buckets by vintage, investor owned properties do prepay a bit more slowly over time, but mimic the shape of the prepayment curve of owner-occupied properties. Below are graphs by vintage in current-coupon buckets of occupancy data. Exhibit 17: Alt-A occupancy prepayment speeds by vintage and WAC
1999 Vintage, 7.75 - 8.75% WAC Bucket 70% 60% 3 Month CPR

2000 Vintage, 8.75 - 9.75% WAC Bucket 70% 60% 3 Month CPR 50% 40% 30% 20% 10%

50% 40% 30% 20% 10% 0% Jan-99 Jan-00 Jan-01 Jan-02 Jan-03 Jan-04

0% Jan-00

Jan-01

Jan-02 Owner

Jan-03 Investor

Jan-04

Owner

Investor

2001 Vintage, 7.50 - 8.50% WAC Bucket 70% 60% 3 Month CPR

2002 Vintage, 6.75 - 7.75% WAC Bucket 70% 60% 3 Month CPR 50% 40% 30% 20% 10%

50% 40% 30% 20% 10% 0% Jan-01 Jan-02 Owner Jan-03 Investor Jan-04

0% Jan-02

Jan-03 Owner Investor

Jan-04

Sources: Loan Performance, Goldman, Sachs & Co.

Goldman Sachs

17

Mortgage Strategies

United States

Another Alt-A characteristic that contributes to prepayment behavior is the documentation of the loan. As would be expected, more than 50% of all loans in our vintage WAC buckets have low or no documentation. As with non-owner occupied properties, this can be a way of achieving higher coupons (to adjust for a lack of documentation) and slower prepays from otherwise prime borrowers (who have fewer refinancing options). Exhibit 18 lists documentation by vintage and WAC. It should be noted that lower WAC buckets have a higher percentage of full documentation loans. Exhibit 18: Alt-A documentation by vintage and WAC bucket Full Documentation 38.9% 40.1% 38.1% 44.6% 34.2% 24.7% 42.7% 34.0% 26.4% 40.6% 38.7% 33.8% Low Documentation 58.5% 55.3% 54.9% 54.6% 61.4% 65.0% 55.4% 61.4% 58.9% 55.9% 56.5% 51.6% No Documentation 2.7% 4.6% 7.0% 0.9% 4.4% 10.3% 1.9% 4.6% 14.7% 3.5% 4.7% 14.6%

1999

2000

2001

2002

< 7.75% Bucket 7.75 - 8.75% Bucket > 8.75% Bucket < 8.75% Bucket 8.75 - 9.75% Bucket > 9.75% Bucket < 7.5% Bucket 7.5 - 8.5% Bucket > 8.5% Bucket < 6.75% Bucket 6.75 - 7.75% Bucket > 7.75% Bucket

Sources: Loan Performance, Goldman, Sachs & Co.

18

Goldman Sachs

United States

Mortgage Strategies

A look at the entire Alt-A universe shows that historically, documentation has little effect on prepayments when examined in a vacuum. Theoretically it would make sense that borrowers with low or no documentation would prepay more slowly, since some impairment would probably make refinancing more difficult. Nonetheless, they could possibly refinance into other Alt-A type loans as rates drop, or obtain previously missing documentation when refinancing. In addition, in order to reduce their exposure to low or no documentation, many originators enforce stricter underwriting guidelines on loans with low documentation, such as lower 4 loan balances, higher FICOs, and lower LTVs . These characteristics that are typically coupled with low and no documentation loans produce better convexity. Exhibit 19: Alt-A documentation prepayment speeds by vintage and WAC
1999 Vintage, 7.75 - 8.75% WAC Bucket 70% 60% 3 Month CPR 3 Month CPR 50% 40% 30% 20% 10% 0% Jan-99 Jan-00 Jan-01 Full Jan-02 None Jan-03 Low Jan-04 70% 60% 50% 40% 30% 20% 10% 0% Jan-00 Jan-01 Jan-02 Full None Jan-03 Low Jan-04 2000 Vintage, 8.75 - 9.75% WAC Bucket

2001 Vintage, 7.50 - 8.50% WAC Bucket 70% 60% 3 Month CPR

2002 Vintage, 6.75 - 7.75% WAC Bucket 70% 60% 3 Month CPR 50% 40% 30% 20% 10%

50% 40% 30% 20% 10% 0% Jan-01 Jan-02 Full Jan-03 None Low Jan-04

0% Jan-02

Jan-03 Full None

Jan-04 Low

Sources: Loan Performance, Goldman, Sachs & Co.

Please see Appendix B for documentation risk layering.

19

Goldman Sachs

Mortgage Strategies

United States

As would also be expected, prepayment penalties are more likely on higher coupon loans, because such loans involve lower quality borrowers. In addition, higher coupon loans are more likely to carry insurance. Over the last several years, the total amount of loans with prepayment penalties has risen, while the amount of loans with insurance has stayed about the same. This is a clear sign that originators have expanded down the credit spectrum. Exhibit 20: Alt-A prepayment penalties and primary mortgage insurance by vintage and WAC buckets Prepayment Penalty (% Yes) 7.6% 8.4% 14.2% 16.8% 20.8% 37.4% 15.3% 16.6% 32.5% 14.3% 15.3% 31.9% Primary Mortgage Insurance (% Yes) 12.2% 24.0% 46.6% 19.1% 38.7% 54.9% 13.7% 26.1% 45.9% 10.4% 20.7% 38.5%

1999

< 7.75% Bucket 7.75 - 8.75% Bucket > 8.75% Bucket

2000

< 8.75% Bucket 8.75 - 9.75% Bucket > 9.75% Bucket

2001

< 7.5% Bucket 7.5 - 8.5% Bucket > 8.5% Bucket

2002

< 6.75% Bucket 6.75 - 7.75% Bucket > 7.75% Bucket

Sources: Loan Performance, Goldman, Sachs & Co.

20

Goldman Sachs

United States

Mortgage Strategies

Exhibit 21 shows that loans with prepayment penalties have historically prepaid more slowly, as expected. Investors are less likely to prepay their mortgage if they have to pay a penalty in order to do so. Exhibit 21: Alt-A prepayment penalties speeds by Vintage and WAC buckets
1999 Vintage, 7.75 - 8.75% WAC Bucket Prepayment Penalty 70% 60% 3 Month CPR
3 Month CPR 70% 60% 50% 40% 30% 20% 10% 2000 Vintage, 8.75 - 9.75% WAC Bucket Prepayment Penalty

50% 40% 30% 20% 10% 0% Jan-99 Jan-00 Jan-01 PP Jan-02 No PP Jan-03 Jan-04

0% Jan-00

Jan-01

Jan-02 PP

Jan-03 No PP

Jan-04

2001 Vintage, 7.50 - 8.50% WAC Bucket Prepayment Penalty 70% 60% 3 Month CPR
3 Month CPR 70% 60% 50% 40% 30% 20% 10%

2002 Vintage, 6.75 - 7.75% WAC Bucket Prepayment Penalty

50% 40% 30% 20% 10% 0% Jan-01 Jan-02 PP Jan-03 No PP Jan-04

0% Jan-02

Jan-03 PP No PP

Jan-04

Sources: Loan Performance, Goldman, Sachs & Co.

Goldman Sachs

21

Mortgage Strategies

United States

Historically, PMI does not seem to be an indicator of prepayments. This measure becomes important when examining the credit of different Alt-A mortgages, and insurance will provide benefits for high LTV defaulted loans. Exhibit 22: Alt-A PMI speeds by vintage and WAC buckets
1999 Vintage, 7.75 - 8.75% WAC Bucket PMI 70% 60% 3 Month CPR

2000 Vintage, 8.75 - 9.75% WAC Bucket PMI 70% 60% 3 Month CPR 50% 40% 30% 20% 10%

50% 40% 30% 20% 10% 0% Jan-99 Jan-00 Jan-01 PMI Jan-02 Jan-03 Jan-04

0% Jan-00

Jan-01

Jan-02 PMI

Jan-03 No PMI

Jan-04

No PMI

2001 Vintage, 7.50 - 8.50% WAC Bucket PMI 70% 60% 3 Month CPR

2002 Vintage, 6.75 - 7.75% WAC Bucket PMI 70% 60% 3 Month CPR 50% 40% 30% 20% 10%

50% 40% 30% 20% 10% 0% Jan-01 Jan-02 PMI Jan-03 No PMI Jan-04

0% Jan-02

Jan-03 PMI No PMI

Jan-04

Sources: Loan Performance, Goldman, Sachs & Co.

22

Goldman Sachs

United States

Mortgage Strategies

A main driver of voluntary refinancing is the ability to do so. One way that this is captured is through a spread at origination, or SATO, analysis. This value for each loan is calculated as the difference between the loan rate and the rate on a par-priced FNMA 30-year mortgage for each period. The different SATO calculations are then bucketed to create prepayment history. While all CPR curves generally follow the interest rate environment, the loans with SATO equal to 0 have the slowest prepays and the loans with the largest spread have the fastest prepays. Exhibit 23: Alt-A SATO analysis
60% 50% 40% CPR 30% 20% 10% 0% Feb-98

Feb-99 <=0%

Feb-00

Feb-01

Feb-02 (0.75, 1.5]

Feb-03

Feb-04

(0, 0.75]

>1.5%

Sources: Loan Performance, Goldman, Sachs & Co.

When we compare these results with FNMA (prime) current coupon loans, we observe a similar pattern, even though the FNMA collateral is all conforming. FICO scores have continued to rise over time, and LTV ranges have stayed relatively stagnant. Exhibit 24: FNMA current coupon characteristics by vintage and WAC bucket Average Loan Size Credit Score Owner Occupied Second Home Investor Owned
1.7% 3.8% 7.3% 2.1% 4.7% 11.7% 1.8% 5.1% 17.6% 1.7% 6.4% 9.3%

1999 2000 2001 2002

Coupon

6.0 6.5 7.0 7.0 7.5 8.0 6.0 6.5 7.0 5.5 6.0 6.5

110,547 101,774 98,246 102,857 94,806 84,183 131,642 120,138 104,058 161,368 142,312 127,269

718 709 698 697 690 682 716 703 688 728 715 701

LTV

75 76 78 79 80 80 75 77 79 71 74 77

96.0% 93.1% 89.5% 94.6% 91.0% 84.5% 95.7% 91.5% 79.5% 95.7% 90.5% 87.2%

2.3% 3.1% 3.2% 3.3% 4.3% 3.7% 2.5% 3.3% 3.0% 2.6% 3.1% 3.5%

Sources: Loan Performance, Goldman, Sachs & Co.

Goldman Sachs

23

Mortgage Strategies

United States

While historically Alt-A loans have mainly been fixed, a trend of originating ARMs has emerged. As we noted in the first section of this paper, the industry has gone from producing mostly fixed loans to producing close to 70% with some adjustable component. Interestingly, the type of adjustable rate issued has shifted over time. More recently, 3/1, 3/6, 5/1, and 5/6 hybrids have been the products of choice. Appendix A lists characteristics of adjustable rate mortgages. As would be expected, CPR performance is fairly constant over product classes. Shorter fixed-period hybrids have a slight tendency to prepay a bit faster than hybrids with longer fixed periods.

24

Goldman Sachs

United States

Mortgage Strategies

Exhibit 25: Hybrid Alt-A prepayments (categories with over $1m outstanding)
Origination Year 1996 1997 Loan Type ARM (1/1) ARM (1/1) 2/1 5/1 ARM (1/1) 2/1 5/1 7/1 10/1 ARM (1/1) 2/1 3/1 5/1 7/1 ARM (1/1) 2/1 3/1 5/1 7/1 10/1 ARM (1/1) 2/1 3/1 5/1 7/1 10/1 ARM (1/1) 2/1 3/1 5/1 7/1 10/1 ARM (1/1) 2/1 3/1 5/1 7/1 10/1 ARM (1/1) 2/1 3/1 5/1 7/1 10/1 Active Investor Balance ($) 5,302,474 3,114,049 4,681,689 5,283,782 5,128,877 3,794,942 17,706,619 4,147,933 2,834,074 4,444,638 58,943,232 10,102,595 35,657,739 11,047,072 11,874,321 36,358,681 18,579,518 66,331,964 8,194,898 4,368,369 262,535,666 276,401,806 136,235,058 531,070,296 94,185,919 15,240,080 1,107,638,719 664,948,783 395,408,398 1,984,656,536 407,791,466 55,033,790 644,933,872 3,936,889,848 4,041,707,910 10,961,841,645 950,102,949 356,452,901 211,045,918 3,296,176,567 5,004,249,366 7,408,806,855 564,281,326 269,346,267 1 Month CPR 33% 27% 11% 40% 1% 7% 56% N/A N/A 98% 55% 29% 67% 72% 50% 67% 18% 57% 41% N/A 46% 55% 51% 64% 50% 39% 53% 69% 59% 52% 32% N/A 40% 43% 44% 37% 13% 17% 29% 17% 21% 15% 11% 8% 3 Month CPR 22% 29% 20% 46% 25% 15% 43% 41% 53% 79% 50% 37% 56% 52% 52% 54% 53% 52% 85% 68% 54% 59% 55% 63% 61% 48% 48% 66% 59% 57% 39% 20% 39% 39% 40% 39% 24% 23% 25% 15% 19% 15% 7% 15% 6 Month CPR 22% 26% 23% 42% 20% 35% 40% 69% 32% 55% 49% 36% 50% 40% 39% 50% 53% 53% 72% 73% 49% 63% 58% 57% 58% 61% 43% 61% 55% 54% 38% 22% 34% 32% 35% 36% 22% 20% 17% 9% 14% 36% N/A N/A 12 Month CPR N/A 12% 0% 34% 19% 39% 41% 65% 59% 44% 47% 54% 54% 46% 26% 67% 54% 61% 72% 73% 46% 64% 62% 61% 62% 60% 41% 55% 55% 53% 40% 22% 33% 34% 40% 37% 23% 20% N/A N/A N/A N/A N/A N/A

1998

1999

2000

2001

2002

2003

2004

Sources: Loan Performance, Goldman, Sachs & Co.

Goldman Sachs

25

Mortgage Strategies

United States

Refinancing incentive
Below we move into the credit aspect of the Alt-A universe and examine delinquencies and defaults. But first lets look at S-curves, which relate refinancing incentive to prepayments. As would be expected, the S-curve for Alt-A collateral has the same general shape as that of jumbo and agency collateral, yet with less overall convexity. With the same incentive, Alt-A borrowers are less likely to refinance a reflection of the issues that did not allow them to apply as prime borrowers originally. In an environment without rate incentives, Alt-A borrowers are more likely to prepay owing to defaults and credit curing. Exhibit 26: MBS S-curves
100 90 80 70 CPR 60 50 40 30 20 10 0 -2 -1 Alt-A 0 Incentive Jumbo Subprime 1 2 3

Sources: Loan Performance, Intex, Goldman, Sachs & Co.

Below are S-curves for the larger Alt-A issuers compared with the agency FNMA Scurve. Exhibit 27: Alt-A S-curves
60

50

40 1 Month CPR

30

20

10

(3.00) (2.00) (1.00) 1.00 Incentive CountryWide RAST RALI FNM 2.00 3.00 4.00

Sources: Intex, Goldman, Sachs & Co.

26

Goldman Sachs

United States

Mortgage Strategies

As with the historical CPR analysis, it is interesting to look at S-curves for different Alt-A characteristics. As would be expected from the above, occupancy is a determinate factor in the convexity story of Alt-A securities. Investor properties (i.e., those purchased solely as an investment) are clearly less negatively convex than owner-occupied properties, with properties purchased as second homes falling somewhere in the middle. Exhibit 28: Alt-A S-curve by occupancy
60%

50%

40% CPR

30%

20%

10%

-2.0%

-1.5%

-1.0%

-0.5%

0% 0.0%

0.5%

1.0%

1.5%

2.0%

2.5%

3.0%

Spread to FHLMC 30 Year Owner Investor Second

Sources: Loan Performance, Goldman, Sachs & Co.

In theory, loans with low or no documentation should have less negative convexity than loans with full documentation owing to their inability to finance. However, as we observed in the CPR analysis, the S-curves of loans with low versus full documentation show little pickup for convexity. This is again due to the risk layering aspect of higher requirements for low documentation loans. Exhibit 29: Alt-A S-curve by documentation
60%

50%

40% CPR

30%

20%

10%

-2.0%

-1.5%

-1.0%

-0.5%

0% 0.0%

0.5%

1.0%

1.5%

2.0%

2.5%

3.0%

Spread to FHLMC 30 Year Full None Low

Sources: Loan Performance, Goldman, Sachs & Co.

Goldman Sachs

27

Mortgage Strategies

United States

Delinquencies
As we would expect, Alt-A delinquencies fall between those of non-agency prime and subprime loans. Exhibit 30 compares delinquencies for Alt-A securities versus those of other non-agency MBS. Exhibit 30: Non-agency delinquencies
Alt-A 30 - 59 Day Delinquency 10% % of Balance
% of Balance 4% 3% 2% 1% 0%

Subprime

Prime 60 - 89 Day Delinquency

8% 6% 4% 2% 0% 0 6 12 18 24 30 Age
90 + Day Delinquency 20% 18% 16% 14% 12% 10% 8% 6% 4% 2% 0% 0 6 12 18 24 30 Age Foreclosure 8% 7% 6% 5% 4% 3% 2% 1% 0% 0 6 12 18 24 30 Age 36 42 48 54 60 36 42 48 54 60

36

42

48

54

60

12

18

24

30 Age

36

42

48

54

60

Bankruptcy 8% 7% 6% 5% 4% 3% 2% 1% 0% 0 6 12 18 24 30 Age 36 42 48 54 60

% of Balance

% of Balance

REO 4% 3% % of Balance 3% 2% 2% 1% 1% 0% 0 6 12 18 24 30 Age 36 42 48 54 60

Sources: Loan Performance, Goldman, Sachs & Co.

% of Balance

28

Goldman Sachs

United States

Mortgage Strategies

To start our investigation, lets look at 60+ day delinquencies for all Alt-A WAC buckets. For the first three years, the 60+ day delinquencies are directly correlated to the coupon on the loans. Higher coupons, which generally indicate poorer credit quality for similar vintages, have the highest delinquencies. After year 3, the 5.0-6.0% and < 5.0% coupon buckets continue to rise, but the delinquencies remain under 7.0%. Exhibit 31: Alt-A 60+ day delinquencies
25% Delinquency as % of MBA Balance

20%

15%

10%

5%

0% 0 6 12 18 24 30 Age 0+ - 5.00% 7.00%+ - 8.00% 10.00%+ - 11.00%

Sources: Loan Performance, Goldman, Sachs & Co.

36

42

48

54

60

5.00%+ - 6.00% 8.00%+ - 9.00%

6.00%+ - 7.00% 9.00%+ - 10.00%

Goldman Sachs

29

Mortgage Strategies

United States

Loan size, FICO, and LTV show historical delinquencies, as one would expect. Larger loans have high delinquencies, as do loans with lower FICO scores and higher LTVs. Exhibit 32: Alt-A 60+ day delinquencies by loan size, FICO, and LTV
By Loan Size 20% 18% 16% 14% 12% 10% 8% 6% 4% 2% 0% 0 6 12 <300K 18 24 30 Age 300 - 600K 600K - 1M 1M+
< 540 540 - 599

By FICO

45% 40% 35% 30% 25% 20% 15% 10% 5% 0% 0 6 12 18 24 30 Age

600 - 659 660 - 719 > 720

36

42

48

54

60

60+ Days Delinquency Rate

60+ Days Delinquency Rate

36

42

48

54

60

By LTV 60+ Days Delinquency Rate 12% 10% 8% 6% 4% 2% 0% 0 6 < 60 12 18 24 30 Age 60+ - 70 70+ - 80 80+ - 90 > 90 36 42 48 54 60

Sources: Loan Performance, Goldman, Sachs & Co.

30

Goldman Sachs

United States

Mortgage Strategies

For occupancy, delinquencies for owner-occupied properties were higher than they were for second-home or investor properties. For second-home properties, it is probably reasonable to assume that people who can afford a second home would not find themselves having to miss payments and if they did, they might consider selling the second home. People who have investor properties are probably receiving some income from the property, possibly in the form of rent from a third party; thus they also are not likely to miss payments. Exhibit 33: Alt-A 60+ day delinquencies by occupancy
6%

5% 60+ Days Delinquency Rate

4%

3%

2%

1%

0% 0 6 12 18 24 30 Age Owner Second Home Non-Owner/Investor 36 42 48 54 60

Sources: Loan Performance, Goldman, Sachs & Co.

Historically, loans with no documentation have had much higher delinquency rates than those with full or low documentation. Exhibit 34: Alt-A 60+ day delinquencies by documentation
16% 14% 60+ Days Delinquency Rate 12% 10% 8% 6% 4% 2% 0% 0 6 12 18 24 30 Age Full Low None 36 42 48 54 60

Sources: Loan Performance, Goldman, Sachs & Co.

Goldman Sachs

31

Mortgage Strategies

United States

Loans with PMI have historically had higher delinquencies. This might be due to the fact that PMI is required for lower quality borrowers where a higher LTV and other impairing characteristics are involved. Exhibit 35: Alt-A 60+ day delinquencies by PMI
10% 9% 60+ Days Delinquency Rate 8% 7% 6% 5% 4% 3% 2% 1% 0% 0 6 12 18 24 30 Age No PMI PMI 36 42 48 54 60

Sources: Loan Performance, Goldman, Sachs & Co.

Cumulative defaults
As we noted at the beginning of this paper, default levels for Alt-A collateral fall between those of prime and subprime. Cumulative defaults for Alt-A collateral as a whole generally fall under 2%, compared with subprime at closer to 11% and prime at under 0.25%. Exhibit 36: Non-agency cumulative defaults
12%

10% Cumulative Defaults

8%

6%

4%

2%

0% 0 6 12 18 24 30 36 Age Subprime Alt-A Jumbo 42 48 54 60 66 72

Sources: Loan Performance, Goldman, Sachs & Co.

32

Goldman Sachs

United States

Mortgage Strategies

Defaults can change drastically by vintage. Economic conditions play an important role in the ongoing defaults of loans. As Exhibit 37 shows, loans originated in 2000 have historically had the highest defaults, followed by loans originated in 2001. These loans were originated during a time when the economy was booming and many people were making higher sums of money than usual. When the economy slumped, people who had been able to make large mortgage payments in the past were less likely to be able to maintain their previous lifestyle. Exhibit 37: Alt-A cumulative defaults by vintage
3.0%

2.5% Cumulative Defaults

2.0%

1.5%

1.0%

0.5%

0.0% 0 4 8 12 16 20 24 28 32 36 40 44 48 52 56 60 Age 1998 1999 2000 2001 2002 2003

Sources: Loan Performance, Goldman, Sachs & Co.

While ARMs currently represent 70% of the origination market, this is a new trend; hence we do not have enough data to use in comparing fixed rate mortgages and ARMs. Nonetheless, for the initial 18 months, there is no difference in the behavior of these two types of loans.

Goldman Sachs

33

Mortgage Strategies

United States

Cumulative defaults based on loan size do not show significant differences. In the first four years, the higher the loan amount, the higher the defaults. However, after the four-year mark is reached, an inflection is observed and lower-balance loans have higher cumulative defaults. This has do with the fact that borrowers with bad credit can obtain only small loans; thus, over time, loans for borrowers in this category tend to have higher defaults. Exhibit 38: Alt-A cumulative defaults by loan size
2.5%

2.0% Cumulative Defaults

1.5%

1.0%

0.5%

0.0% 0 4 8 <=50k 12 16 20 24 28
Age

32

36

40

44

48

52

56 >200k

60

(100k, 150k]

(150k, 200k]

(50k, 100k]

Sources: Loan Performance, Goldman, Sachs & Co.

As would be expected, loans with higher FICO scores have lower defaults, as the FICO score is a direct indicator of credit performance. Exhibit 39: Alt-A cumulative defaults by FICO
14% 12% 10% 8% 6% 4% 2% 0% 0 4 8 12 16 20 24 28 32 36 40 44 48 52 56 60 Age <550 [550,600) [600,650) [650,700) >700

Sources: Loan Performance, Goldman, Sachs & Co.

Cumulative Defaults

34

Goldman Sachs

United States

Mortgage Strategies

We observe a similar pattern when we look at loan buckets of LTVs. The lower the LTV, the lower the cumulative default rate. Besides indicating that the borrowers are of higher quality, lower LTVs mean the borrowers have greater equity in their homes, and hence are less likely to default. It is interesting to note that loans with LTVs greater than 100 have a cumulative default rate of 8%, while loans with LTVs between 80 and 100 have cumulative defaults of just over 2%. Exhibit 40: Alt-A cumulative defaults by LTV
9% 8% 7% Cumulative Defaults 6% 5% 4% 3% 2% 1% 0% 0 4 8 12 16 20 24 28 32 36 40 44 48 52 56 60 Age <60 [60,80) [80,100) >=100

Sources: Loan Performance, Goldman, Sachs & Co.

Some Alt-A characteristics have interesting trends with respect to cumulative defaults. As is true of prepayments and delinquencies, occupancy appears to be a good indicator of cumulative defaults. Owner occupied houses have the highest rate of default, followed by investment properties and then by properties that are second homes. This makes sense: Investors in real estate as well as those who can afford a second home are likely to be in a more stable financial position than people who are taking out a mortgage on their only home. Exhibit 41: Alt-A cumulative defaults by occupancy
2.0% 1.8% 1.6% Cumulative Defaults 1.4% 1.2% 1.0% 0.8% 0.6% 0.4% 0.2% 0.0% 0 4 8 12 16 20 24 28 32 36 40 44 48 52 56 60 Age Non-Owner Owner 2nd Home

Sources: Loan Performance, Goldman, Sachs & Co.

Goldman Sachs

35

Mortgage Strategies

United States

At first glance, documentation does not seem to be an indicator of defaults. However, as with prepayments, the risk of many loans with low documentation is mitigated by low LTVs and higher FICO scores, hence limiting the poor performance of these loans. Exhibit 42: Alt-A cumulative defaults by documentation
2.0% 1.8% 1.6% Cumulative Defaults 1.4% 1.2% 1.0% 0.8% 0.6% 0.4% 0.2% 0.0% 0 4 8 12 16 20 24 28 32 36 40 44 48 52 56 60 Age Full Low/reduced

Sources: Loan Performance, Goldman, Sachs & Co.

36

Goldman Sachs

United States

Mortgage Strategies

Severity
Severities are the next performance measure to examine in connection with defaults. Once a loan defaults, it no longer matters if it was originally classified as a subprime, Alt-A, or prime. Instead, the important characteristics are the LTV and size of the loan. When looking at different Alt-A characteristics in this space, LTV and loan size are still the drivers of the severity, but occupancy also seems to play a role. The severity of loans of investor properties is almost double that of owner-occupied properties or second homes. In general, investor properties have lower LTVs and lower loan balances. The sharp increase in severities in investor properties reflects two characteristics. First, when such a loan defaults, the investor makes less of an effort to preserve the property, which causes a decline in its value. Second, residential investment properties are typically in less desirable locations and experience more-volatile market price movement. Exhibit 43: Alt-A severities by characteristic
Vintage Severity Loan Size Severity LTV Severity FICO Severity Occupancy Type Severity Doc Type Severity Purpose Severity 1998 27.68% <100k 42.83% <90 30.64% <550 36.38% 2ndHome 27.58% Full 30.78% Cashout 30.04% 1999 37.55% [100k,150k) 29.41% [90,100) 25.38% [550,600) 40.41% Investor 40.54% Low 28.13% Purchase 29.25% 2000 32.05% [150k,300k) 23.09% >100 32.52% [600,650) 31.07% Owner 27.12% NoDoc 28.94% Rate/Term 29.21% 2001 22.51% >=300k 27.51% 2002 24.84% 2003 23.16%

[650,700) 31.03%

>700 28.75%

Sources: Loan Performance, Goldman, Sachs & Co.

Goldman Sachs

37

Mortgage Strategies

United States

Modeling Alt-A
In order to model any new type of credit mortgage collateral, adjustments are made to base prepayment and default curves. These changes are referred to as the relative strength of different characteristics, and are developed through historical information and the views of the Goldman Sachs mortgage strategy group. Exhibit 44 provides relative strength factors for different Alt-A collateral using the following prime base case: LTV between 70 and 80, loan size between $250,000 and $350,000, FICO greater than 739, full documentation, and owner occupied. The characteristics that are the most determinant for prepayments (i.e., that have the greatest effect on prepayments) are loan size and occupancy. Loan size causes the curve to shift by a 50% slowdown for small loan balances and a 35% speed increase for larger balances. Investor owned properties cause the curve to shift 30% slower. Exhibit 44: Prepayment relative strength Characteristic
720 < fico score <= 739 679 < fico score <= 720 639 < fico score <= 679 200 < fico score <= 639 Loan size < 50k 50k < loan size < 150k 150k < loan size < 250k 350k < loan size < 450k 450k < loan size < 550k Loan size > 550k 10 <= ltv < 70 80 <= ltv < 90 90 <= ltv Low Documentation Non-Owner Occupied
Source: Goldman, Sachs & Co.

Fraction of agency default curve 15yr

0.955 0.951 0.920 0.897 0.493 0.617 0.779 1.215 1.319 1.367 1.085 0.989 0.991 0.951 0.793

30yr
0.980 0.943 0.868 0.824 0.495 0.620 0.798 1.198 1.279 1.285 1.085 0.989 0.991 0.965 0.884

38

Goldman Sachs

United States

Mortgage Strategies

When looking at the relative strength from a default perspective, the most determinant characteristics are LTV and documentation. In analyzing the cumulative default curve, we can obtain factors for similar characteristics. In this example, we use a base cumulative default curve of 80 LTV, 710 FICO, $100,000 loan size, full documentation, and owner occupied. Unlike in the case of prepayments, LTV has the greatest fluctuation in relative strength, with defaults increasing as much as 500% for high LTVs and defaults decreasing as much as 90% for low LTVs. Documentation also has a large impact on defaults. Exhibit 45: Default relative strength Characteristic
FICO = 640 FICO = 680 FICO = 750 FICO = 780 Loan Size = 50,000 Loan Size = 70,000 Loan Size = 180,000 Loan Size = 290,000 Loan Size = 400,000 LTV = 55 LTV = 65 LTV = 75 LTV = 90 LTV = 100 Low Documentation Non-Owner Occupied
Source: Goldman, Sachs & Co.

Fraction of agency prepayment curve

30yr

3.194 2.013 0.662 0.436 1.914 1.273 0.908 1.594 2.346 0.071 0.309 0.869 1.499 4.582 1.806 1.018

Conclusion
While the analysis throughout this report would suggest that deals with a high concentration of investor property have better convexity, lower delinquencies, and lower defaults, Alt-A collateral is complex, with many characteristics contributing to overall performance. As the Alt-A market expands and this type of collateral is better understood, participating in the market becomes more attractive. Thus we expect to see new classifications of Alt-A characteristics. The next step in this analysis would be to examine how different risk factors are layered in Alt-A loans and securities..

Goldman Sachs

39

Appendix A: Alt-A Adjustable Rate Mortgage Characteristics Origination Year Orig. Number of Loans Avg Loan Size Orig. WA Credit Score Orig. WA LTV NonOwner/ Investor PP Penalty %

1996

ARM (1/1) 2/1 3/1 5/1 7/1

Loan Type

Original Investor Balance

3,871,816,435 43,348,869

20,099 544 6,335 139 166 593 897 124 657 138 31 1,002 3,015 72 1,789 913 215 682

192,637 79,685 203,640 180,932 256,713 104,320 96,003 100,577 351,918 238,935 374,942 210,435 163,734 547,250 391,827 456,801 542,170 192,220 131,262 193,854 314,122 143,976 342,492 516,989 173,839 324,924 386,618 179,925 606,661

694

74

Owner Occ.

82.2%

Second Home

6.6%

11.2% 9.7% 4.8% 2.8% 0.0% 4.7% 6.0% 5.4% 0.0% 0.0% 0.0% 5.4% 0.0% 0.0% 0.4% 0.5% 0.0% 2.2% 1.3% 8.9%

Full Doc

43.8%

Low Doc

56.2%

No Doc

0.0%

CA %

33.8%

0.7%

PMI %

11.7% 0.0%

615 693 641 689 622 613 669 705 708 717 689 669 710 725 710 650 687 631 675 710 727 757 690 653 709 714 728 714

73 74 65 79 82 80 87 77 89 55 83 88 66 71 73 69 78 80 75 71 95 81 66 83 68 72 91 69

86.7% 84.4% 97.2% 100.0% 92.3% 89.8% 94.6% 100.0% 100.0% 100.0% 70.3% 98.5% 100.0% 94.5% 99.5% 100.0% 97.8% 98.7% 89.5% 72.9% 95.6% 75.4% 95.3% 97.2% 80.3% 75.6% 92.4% 87.6%

3.6% 10.8% 0.0% 0.0% 3.0% 4.2% 0.0% 0.0% 0.0% 0.0% 24.3% 1.5% 0.0% 5.1% 0.0% 0.0% 0.0% 0.1% 1.6% 9.3% 0.7% 24.6% 2.8% 0.8% 6.7% 7.4% 3.7% 8.9%

53.1% 37.2% 100.0% 100.0% 33.0% 17.9% 59.5% 84.1% 100.0% 100.0% 32.8% 3.8% 9.1% 94.9% 36.9% 90.4% 26.7% 85.0% 56.2% 27.4% 88.5% 87.4% 32.4% 53.9% 30.2% 33.1% 82.2% 44.6%

46.9% 62.8% 0.0% 0.0% 67.0% 81.8% 40.5% 15.9% 0.0% 0.0% 67.2% 96.2% 0.0% 3.7% 52.8% 9.6% 68.1% 10.4% 31.5% 54.0% 9.1% 12.6% 62.6% 29.7% 56.0% 54.0% 14.4% 48.4%

0.0% 0.0% 0.0% 0.0% 0.0% 0.3% 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 90.9% 1.4% 10.3% 0.0% 5.3% 4.6% 12.3% 18.7% 2.4% 0.0% 5.0% 16.4% 13.9% 12.9% 3.5% 7.1%

41.2% 55.1% 0.0% 19.5% 18.6% 29.7% 0.0% 41.8% 26.3% 100.0% 16.3% 53.9% 9.1% 21.4% 59.7% 21.3% 8.2% 44.3% 63.6% 55.4% 20.0% 0.0% 53.2% 50.5% 57.7% 56.9% 26.8% 42.7%

59.2% 0.0% 0.0% 0.0% 30.9% 64.4% 0.0% 0.0% 0.0% 0.0% 42.5% 76.1% 0.0% 0.0% 10.3% 0.0% 32.4% 91.2% 60.3% 57.9% 14.1% 0.0% 37.6% 51.6% 27.9% 25.4% 22.6% 39.0%

1,290,059,180 25,149,600 42,614,400 61,861,750 86,114,455 12,471,525 231,209,926 32,972,995 11,623,200 210,856,209 493,657,384 39,402,000 700,979,221 417,059,635 116,566,538 131,094,350 15,904,215,536 5,318,184,322 10,108,129,938 3,078,936,691 48,976,381 468,391,710 3,035,403,766 6,475,413,177 20,284,680,117 8,002,328,285 1,071,970,179

13.7% 32.4% 25.4% 40.4% 33.0% 82.6% 17.8% 68.3% 0.0% 39.7% 84.3% 0.0% 12.3% 5.1% 3.9% 19.5% 6.1% 5.5% 9.1% 78.9% 0.0% 4.1% 34.1% 7.9% 19.0% 69.8% 1.1%

1997

ARM (1/1) 2/1 3/1 5/1 7/1 10/1

1998

ARM (1/1) 2/1 3/1 5/1 7/1 10/1

1999

ARM (1/1) 2/1 3/1 5/1 7/1 10/1

121,164 27,434 32,179 21,385 143 906 17,461 19,929 52,467 44,476 1,767

17.8% 3.7% 0.0% 1.9% 2.0% 13.0% 16.9% 3.9% 3.5%

2000

ARM (1/1) 2/1 3/1 5/1 7/1 10/1

Appendix A: Alt-A Adjustable Rate Mortgage Characteristics Origination Year

2001

Loan Type
ARM (1/1) 2/1 3/1 5/1 7/1 10/1

Original Investor Balance

28,288,972,477 47,824,079,098 23,279,007,570 100,486,212,478 2,236,802,702 1,643,783,296 62,458,296,137 43,420,104,638 31,213,143,623 137,755,877,664 18,157,541,050 1,326,512,313 10,663,160,308 52,429,519,521 52,356,638,100 159,980,458,043 12,506,322,481 3,923,420,219 1,930,310,104 20,307,807,975 32,487,033,903 48,551,941,003 4,293,186,472 1,703,747,770

Orig. Number of Loans

87,155 206,879 56,396 220,551 4,766 3,329 224,007 187,145 86,032 375,319 37,566 2,556 37,866 221,434 171,187 470,960 31,635 10,481 6,468 85,682 120,388 185,538 15,767 5,137

Avg Loan Size

324,582 231,169 412,778 455,614 469,325 493,777 278,823 232,013 362,809 367,037 483,350 518,980 281,603 236,773 305,845 339,690 395,332 374,336 298,440 237,014 269,853 261,682 272,289 331,662

Orig. WA Credit Score

695 681 706 717 717 723 693 683 712 716 728 739 693 688 706 713 725 725 689 686 707 712 723 727

Orig. WA LTV
76 82 70 69 71 72 80 82 73 71 65 64 77 81 76 73 66 65 75 82 77 75 73 68

Owner Occ.
89.3% 92.1% 86.6% 83.4% 93.8% 85.7% 85.8% 88.2% 88.7% 83.5% 85.8% 88.7% 79.0% 85.0% 86.2% 83.6% 82.5% 86.4% 80.1% 84.6% 79.4% 80.0% 83.9% 84.1%

Second Home
3.1% 1.0% 4.0% 6.4% 5.6% 6.6% 3.1% 2.2% 3.6% 4.5% 6.1% 2.8% 3.2% 1.8% 2.9% 3.8% 6.2% 7.2% 2.2% 1.9% 4.3% 3.9% 6.7% 5.6%

NonOwner/ Investor
7.5% 6.9% 9.4% 10.3% 0.6% 7.7% 11.1% 9.6% 7.7% 11.9% 8.1% 8.5% 17.8% 13.3% 10.8% 12.6% 11.2% 6.4% 17.7% 13.4% 16.3% 16.2% 9.4% 10.3%

Full Doc
14.3% 13.4% 27.6% 39.1% 79.1% 66.9% 14.6% 16.7% 35.4% 35.2% 41.1% 56.5% 17.2% 17.8% 27.0% 31.1% 42.4% 54.7% 12.5% 27.5% 30.5% 36.9% 46.2% 42.4%

Low Doc
73.7% 76.0% 60.9% 51.0% 12.2% 14.4% 76.7% 74.2% 58.7% 51.0% 41.7% 28.5% 73.6% 74.0% 65.7% 58.1% 40.8% 29.1% 81.8% 64.8% 62.3% 54.0% 46.2% 47.3%

No Doc
12.0% 10.7% 11.5% 9.8% 8.6% 18.7% 8.7% 9.1% 5.9% 13.8% 17.1% 14.9% 9.1% 8.2% 7.3% 10.8% 16.7% 16.2% 5.7% 7.7% 7.2% 9.1% 7.6% 10.3%

CA %
69.5% 62.7% 55.9% 57.2% 26.5% 21.8% 65.9% 56.6% 53.7% 56.1% 47.8% 39.2% 58.8% 60.6% 52.9% 57.7% 45.6% 46.3% 64.7% 54.8% 50.0% 46.2% 42.1% 50.9%

PP Penalty %
42.0% 70.0% 27.6% 23.2% 0.1% 8.9% 71.3% 73.6% 23.8% 31.6% 39.4% 33.0% 67.1% 76.3% 32.5% 28.7% 37.5% 18.1% 69.1% 71.6% 29.3% 24.2% 14.1% 26.3%

PMI %
36.1% 38.6% 9.0% 4.9% 18.2% 5.9% 39.5% 36.5% 8.3% 7.7% 3.5% 1.5% 25.0% 24.8% 9.8% 7.5% 4.4% 3.3% 16.8% 20.0% 10.0% 9.0% 5.5% 2.4%

2002

ARM (1/1) 2/1 3/1 5/1 7/1 10/1

2003

ARM (1/1) 2/1 3/1 5/1 7/1 10/1

2004

ARM (1/1) 2/1 3/1 5/1 7/1 10/1

Appendix B: Top 50 Documentation Combinations Rank

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46

Loan Size Top Level

300k+ - 600k 300k+ - 600k 0+ - 300k 0+ - 300k 0+ - 300k 300k+ - 600k 0+ - 300k 300k+ - 600k 0+ - 300k 300k+ - 600k 0+ - 300k 600k+ - 1M 600k+ - 1M 300k+ - 600k 300k+ - 600k 0+ - 300k 0+ - 300k 600k+ - 1M 600k+ - 1M 0+ - 300k 0+ - 300k 300k+ - 600k 0+ - 300k 300k+ - 600k 0+ - 300k 0+ - 300k 600k+ - 1M 300k+ - 600k 0+ - 300k 300k+ - 600k 0+ - 300k 0+ - 300k 300k+ - 600k 1M+ 0+ - 300k 300k+ - 600k 300k+ - 600k 600k+ - 1M 0+ - 300k 0+ - 300k 600k+ - 1M 300k+ - 600k 300k+ - 600k 0+ - 300k 300k+ - 600k 1M+

FICO
720 - 779 660 - 719 660 - 719 600 - 659 720 - 779 720 - 779 600 - 659 600 - 659 540 - 599 660 - 719 660 - 719 720 - 779 720 - 779 780 - 839 720 - 779 660 - 719 660 - 719 660 - 719 660 - 719 720 - 779 600 - 659 660 - 719 600 - 659 600 - 659 480 - 539 540 - 599 720 - 779 720 - 779 720 - 779 660 - 719 540 - 599 720 - 779 780 - 839 720 - 779 660 - 719 600 - 659 540 - 599 720 - 779 780 - 839 720 - 779 660 - 719 720 - 779 780 - 839 480 - 539 660 - 719 660 - 719

LTV
70+ - 80 70+ - 80 70+ - 80 70+ - 80 70+ - 80 60+ - 70 80+ - 90 70+ - 80 70+ - 80 60+ - 70 80+ - 90 70+ - 80 60+ - 70 70+ - 80 50+ - 60 60+ - 70 90+ - 100 70+ - 80 60+ - 70 60+ - 70 90+ - 100 80+ - 90 60+ - 70 80+ - 90 70+ - 80 80+ - 90 50+ - 60 40+ - 50 90+ - 100 50+ - 60 60+ - 70 80+ - 90 60+ - 70 60+ - 70 50+ - 60 60+ - 70 70+ - 80 40+ - 50 70+ - 80 50+ - 60 50+ - 60 80+ - 90 50+ - 60 60+ - 70 90+ - 100 60+ - 70

Appendix B: Top 50 Documentation Combinations Rank

47 48 49 50

Loan Size Top Level

300k+ - 600k 600k+ - 1M 300k+ - 600k 0+ - 300k

FICO
660 - 719 780 - 839 720 - 779 600 - 659

LTV
40+ - 50 70+ - 80 30+ - 40 50+ - 60

Copyright 2005. The Goldman Sachs Group, Inc. All rights reserved. This material was prepared by the Mortgage Strategies Group within the Fixed Income, Currency and Commodities Division and is not the product of Goldman Sachs Global Investment Research. It is not a research report and should not be construed as such. The Goldman Sachs Group, Inc. is a full-service, integrated investment banking, investment management, and brokerage firm. We are a leading underwriter of securities and a leading participant in virtually all trading markets. We have investment banking and other business relationships with a substantial number of companies. Investors should assume that The Goldman Sachs Group, Inc. and/or its affiliates are conducting, seeking or will seek investment banking or other business from any company or companies that may be mentioned in this material and trade or will trade as principal in the securities of those companies. The professionals who prepared this material are paid in part based on the profitability of The Goldman Sachs Group, Inc., which includes earnings from the firms trading, capital markets, investment banking and other business. They, along with other salespeople, traders, and other professionals may provide oral or written market commentary or trading strategies to our clients that reflect opinions that are contrary to the opinions expressed herein, and our proprietary trading and investing businesses may make investment decisions that are inconsistent with the views expressed herein. In addition, the professionals who prepared this material may also produce material for, and from time to time, may advise, Goldman Sachs sales and trading desks. This material should not be construed as an offer to sell or the solicitation of an offer to buy any security in any jurisdiction where such an offer or solicitation would be illegal. We are not soliciting any action based on this material. It is for the general information of clients of The Goldman Sachs Group, Inc. It does not constitute a personal recommendation or take into account the particular investment objectives, financial situations, or needs of individual clients. Before acting on any advice or recommendation in this material, clients should consider whether it is suitable for their particular circumstances and, if necessary, seek professional advice. The price and value of the investments referred to in this material and the income from them may go down as well as up, and investors may realize losses on any investments. Past performance is not a guide to future performance. Future returns are not guaranteed, and a loss of original capital may occur. The Goldman Sachs Group, Inc. does not provide tax advice to its clients, and all investors are strongly advised to consult with their tax advisers regarding any potential investment. Certain transactions - including those involving futures, options, and other derivatives as well as non-investment-grade securities - give rise to substantial risk and are not suitable for all investors. The material is based on information that we consider reliable, but we do not represent that it is accurate, complete and/or up to date, and it should not be relied on as such. Opinions expressed are our current opinions as of the date appearing on this material only. We and our affiliates, officers, directors, and employees, including persons involved in the preparation or issuance of this material, may from time to time have long or short positions in, act as principal in, and buy or sell the securities or derivatives (including options) thereof in, and act as market maker or specialist in, and serve as a director of, companies mentioned in this material. In addition, we may have served as manager or co-manager of a public offering of securities by any such company within the past three years. This material is distributed in the United States by Goldman, Sachs & Co., in Hong Kong by Goldman Sachs (Asia) L.L.C., in Korea by Goldman Sachs (Asia) L.L.C., Seoul Branch, in Japan by Goldman Sachs (Japan) Ltd., in Australia by Goldman Sachs Australia Pty Limited (ACN 092 589 770), and in Singapore by Goldman Sachs (Singapore) Pte. This material has been issued by The Goldman Sachs Group, Inc. and/or one of its affiliates and has been approved for the purposes of section 21 of the Financial Services and Markets Act 2000 by Goldman Sachs International, which is regulated by the Financial Services Authority, in connection with its distribution in the United Kingdom, and by Goldman Sachs Canada, in

connection with its distribution in Canada. We may, to the extent permitted under applicable law, have acted on or used this material prior to or immediately following its publication. Foreign-currency-denominated securities are subject to fluctuations in exchange rates that could have an adverse effect on the value or price of, or income derived from, the investment. In addition, investors in securities such as ADRs, the values of which are influenced by foreign currencies, effectively assume currency risk. In addition, options involve risk and are not suitable for all investors. Further information on any of the securities mentioned in this material may be obtained on request, and for this purpose, persons in Italy should contact Goldman Sachs S.I.M. S.p.A. in Milan or its London branch office at 133 Fleet Street; persons in Hong Kong should contact Goldman Sachs (Asia) L.L.C. at 2 Queen' Road Central; and persons in Australia should s contact Goldman Sachs Australia Pty Limited. This material is not for distribution in the United Kingdom to private customers, as that term is defined under the rules of The Securities and Futures Authority. A glossary of certain of the financial terms used in this material is also available on request. Unless governing law permits otherwise, you must contact a Goldman Sachs entity in your home jurisdiction if you want to use our services in effecting a transaction in the securities mentioned in this material.