Jp Morgan Cdo Handbook

Global Structured Finance Research CDO Research J.P. Morgan Securities Inc. New York February 19, 2004

Structured Finance CDO Handbook Overview Structured Finance (SF) CDOs are leveraged investment vehicles that invest primarily in the senior and mezzanine tranches of structured products (ABS, RMBS, CMBS, and CDOs). They utilize the same technology as traditional credit CDOs with the only difference being their underlying collateral. SF CDOs issue securities to fund the purchase of collateral or assume risk synthetically via credit derivatives.

Highlights • SF CDOs are designed to exploit arbitrage opportunities by taking advantage of liquidity/complexity premiums and the credit curve, to be a source of funding, or to manage balance sheet exposures. They have been growing as a portion of total CDO issuance.

Contents Overview What’s In an SF CDO? SF CDO Assets: 101 Does SF Collateral “Work” for CDOs? SF CDO Structure Should SF CDOs be Managed or Static? SF CDO Rating Methodologies Appendix A: Rating Transition Matrices Appendix B: Rating Agency Classification of Structured Products Appendix C: SF CDOs from Seasoned Issuers

2 3 12 24 32 44 48 51 53 54

• Overall, the SF CDO underlying collateral makeup largely mirrors the structured products market. Some adjustments are made to enhance arbitrage. • Structured product collateral offers a spread pick-up, lower event risk, and comparable default/recovery rates versus like-rated corporates, as well as diversification opportunities. • Rating agency structured product default and recovery assumptions are conservative compared with actual collateral performance. • SF CDOs have several variables, including quality of collateral (AAA/AA or BBB) and form of exposure (cash or synthetic). Each type has unique structural features. • Manager/issuer selection is critical in both actively managed and static deals. Christopher FlanaganAC

Chart 1

(1-212) 270-6515 [email protected]

Structural Overview of a Typical SF CDO Asset Manager

Hedge

Trustee &

or Issuer

Counterparties

Custodian

Asset Pool

Rishad Ahluwalia (London) (44-207) 777-1045 [email protected] CDO Liabilities

AAA AA

ABS RMBS CMBS CDOs

CDO SPV

A BBB Equity

Source: J.P. Morgan Securities.

The certifying analyst(s) is indicated by a superscript AC. See last page of the report for analyst certification and important legal and regulatory disclosures.

Ryan Asato

(1-212) 270-0317 [email protected]

Benjamin J. Graves

(1-212) 270-1972 [email protected]

Edward Reardon (London)

(44-207) 777-1260 [email protected]

www.morganmarkets.com

February 19, 2004 Analyst Christopher FlanaganAC [email protected]

Global Structured Finance Research CDO Research Structured Finance CDO Handbook

Overview SF CDOs have been growing as a portion of total CDO issuance in both the US (currently about 45% of total) and Europe (currently about 20% of total). SF CDOs are Growing as a Percentage of Total CDO Issuance

Chart 2

Funded CDO Issuance: 1996-2003* US $ Billion (bar graph)

% SF CDO (black line)

Europe $ Billion (bar graph)

45%

80

40%

70

35%

60

30%

50

25%

20

40

20%

15

30

15%

20

10%

10

5% 0%

0 1996 1997 1998 1999 2000 2001 2002 2003

% SF CDO (black line)

35

90

25%

30

20%

25 15%

10%

10 5%

5 0

0% 1996 1997 1998 1999 2000 2001 2002 2003

*Funded issuance includes all cash issuance and the funded (i.e. non super senior) portion of synthetics. Source: JPMS, IFR Markets, MCM, Bloomberg, CreditFlux.

SF CDOs offer a spread pick-up to most like-rated securities. This pick-up can be attributed to liquidity, complexity, and a new asset class premium. SF CDOs Offer a Spread Pick-Up to Most Like Rated Securities

Table 1

Spread to Swaps/Libor (bp)*

AAA AA A BBB

7-12 Yr SF CDO

5-8 Yr IG Syn CDO

6-10 Yr HY CLO

10 Yr RMBS Jumbo

10 Yr CMBS

3-5 Yr HEL

60** 125 175 375

70 125 175 375

46 90 140 265

100 120 175

30 38 45 88

25 50 105 170

10 Yr Floating Cards

18 61 105

UK 5 Yr Sterling 10 yr RMBS Industrial

15 34 107

-8 6 28 60

*As of 1 February 2004. **Indicative weighted average AAA spread for traditional SF CDO. Source: JPMS.

Growing Cadre of Market Participants

This spread pick-up, as well as the opportunity to diversify exposures, is attracting a growing investor base, which varies by position in the SF CDO capital structure. Senior investors include banks, conduits, SIVs, and finance companies. Equity investors are typically banks, pension funds, endowments, private banks, insurance companies, fund managers, and hedge funds. SF CDO asset managers are keen to become involved in this market as a way to increase assets under management, build their franchise, and receive fee income. On the other hand, balance sheet transactions can enable entities to reduce economic and regulatory capital, manage credit risk, and achieve long term funding.

2

February 19, 2004 Analyst Christopher FlanaganAC [email protected]

Global Structured Finance Research CDO Research Structured Finance CDO Handbook

What’s In an SF CDO? SF CDOs source collateral from several distinct sectors of the broader structured products market, including CMBS, RMBS, CDOs, ABS, and REITs. Collateral composition varies by deal, with individual deals sourcing from 0% to 100% of their collateral from each of the above sectors. Several factors, including asset manager experience, issuance volume, region, and arbitrage opportunity influence collateral composition. Despite differences between deals, US/European SF CDOs can be broadly characterized into two sub-sectors1. SF CDOs can be classified as Real Estate or Diversified

Real Estate SF CDO: Greater than 60% of collateral is backed by residential or commercial real estate (RMBS, CMBS, or REITs). In 2003, Real Estate CDOs accounted for approximately 45% of funded SF CDO volume. Diversified SF CDO: Also referred to as “multisector” or “ABS” CDOs. Deals in this category consist of a diversified mix of structured finance assets and, as such, don’t exhibit asset concentration (in Real Estate) described above. In 2003, Diversified CDOs accounted for approximately 55% of funded SF CDO volume. In addition to the two broad categories above, SF CDOs may be distinguished by several other characteristics, including: Cash/Synthetic, Arbitrage/Balance Sheet, or US/Europe. Chart 4 to Chart 9 on the following page show the aggregate collateral composition (Consumer ABS, Commercial ABS, RMBS, CMBS, CDO, REIT) for deals completed in 2003. Chart 10 to Chart 14 detail the collateral breakdown within each of the aforementioned sectors. We stress, however, that different types of SF CDOs hold these underlying asset classes in materially different proportions.

Chart 3

2003 SF CDO Collateral Distribution (total) CDO 15% RMBS 39% CMBS 19%

REIT 3% Other ABS 2%

Corp ABS 6%

Consumer ABS 16%

Source: JPMS, IFR Markets, MCM, S&P, Fitch, Moody’s.

SF CDO Sectors - Some Definitions CDO: CLO, CBO, SF CDO, IG Synthetic CDO, Small-to-Medium Entity CDO, Other CDO CMBS: Conduit, Large Loan, Credit Tenant Lease Consumer ABS: Student Loan, Auto, Card, Consumer Loan Corporate ABS: Equipment, Health Care, Small Business Loan, Structured Settlement, Aircraft, Aerospace, Trade Receivables, Franchise REIT: Unsecured corporate debt of company that invests in Regional Malls, Shopping Centers, Office Buildings, Warehouses RMBS: Prime* (Jumbo, Alt A), Home Equity (Subprime or B/C, 2nd Lien), NIMS, Manufactured Housing, Prime European Mortgages (UK, Netherlands, Spain, Italy, Portugal) * Note: Prime RMBS also traditionally includes the conforming Agency (Fannie Mae, Freddie Mac) paper, which is out of the scope of this paper because it is guaranteed and not typically included in SF CDOs.

1. CDOs-of-CDOs (majority of collateral is tranches of other CDOs) can also be considered SF CDOs, but they are beyond the scope of this paper. 3

February 19, 2004 Analyst Christopher FlanaganAC [email protected]

Global Structured Finance Research CDO Research Structured Finance CDO Handbook

2003 US/Europe SF CDO Collateral Distribution: By Deal Type ($billion notional)* Chart 4

Chart 5

Cash: 16 Euro, 35 US deals ($29.1)

Synthetic: 22 Euro, 4 US deals ($39.8)

Consumer ABS 4%

Corp ABS 4% Corporate 0%

Consumer ABS 28%

RMBS 34%

RMBS 48%

CMBS 33%

Corp ABS 8%

REIT 5%

CMBS 7%

CDO 6%

Chart 6

Arbitrage: 19 Euro, 33 US deals ($32.5) Corp ABS 4% Corporate 0%

CDO 23%

Chart 7

Balance Sheet: 19 Euro, 6 US deals ($36.4)

Consumer ABS 5%

Consumer ABS 27%

RMBS 37%

RMBS 44%

CMBS 31%

Corp ABS 8% REIT 5%

CMBS 9%

CDO 11%

Chart 8

Europe: 38 deals ($43.6)

Consumer ABS 5%

Consumer ABS 23% RMBS 43%

CMBS 34%

REIT 5%

CDO 9%

RMBS 39% Corp ABS 7% Corporate 0% CMBS 11%

Source: JPMS, IFR Markets, MCM, Bloomberg, Moody’s, Standard and Poor’s, Fitch.

* As a percent of total notional value. 4

CDO 19%

Chart 9

US: 39 deals ($25.2) Corp ABS 4% Corporate 0%

REIT 0%

REIT 1%

CDO 19%

2003 US/Europe SF CDO Collateral Distribution: By Sub Sector ($billion notional)* Chart 10

Chart 11

RMBS ($27.4)

CMBS ($13.5) MH 2%

Large Loan 22% Prime 43%

CTL 2%

HEL 55% Conduit 76%

Chart 12

Chart 13

Corp ABS ($4.0)

Consumer ABS ($10.9) Aerospace 1%

Structured Settlement 7%

Student Loan 2%

Aircraft 1%

Consumer Loan 20%

Auto 34%

Equipment 28%

Health Care 2%

Small Business Loan 61%

Card 44%

Chart 14

All

25% 20% 15% 10% 5%

r he Ot

t

MH RM BS :

m ip qu :E

AB S p Co r

m :S AB S

Co

rp

en

T RE I

es in lB us

er um ns Co S:

AB er Co

ns

um

s

an Lo

Lo g :L BS CM

AB S: ns

um

er

AB S: Co

Co

ns

um

er

BS :

an

to Au

rd Ca

CD O

t Co

nd

ui

Pr im e CM

HE L

0% RM BS :

Analyst Christopher FlanaganAC [email protected]

Global Structured Finance Research CDO Research Structured Finance CDO Handbook

RM BS :

February 19, 2004

Source: JPMS, IFR Markets, MCM, Bloomberg, Moody’s, Standard and Poor’s, Fitch.

* As a percent of total notional value. 5

February 19, 2004 Analyst Christopher FlanaganAC [email protected]

SF CDO Collateral Mirrors the Overall Market

Global Structured Finance Research CDO Research Structured Finance CDO Handbook

The SF CDO collateral concentrations shown above are largely representative of the underlying collateral markets as a whole. That is, SF CDOs tend to source collateral from the largest sectors, such as home equities (HELs) and CMBS. The one possible exception (at least with US collateral) is the large prime RMBS market, from which SF CDOs source smaller amounts of collateral. This is due to both the relatively larger rate component of this sector (CDO technology is better equipped to take credit risk than rate risk), as well as the smaller supply of subordinates (due to relatively low credit enhancement associated with the prime quality collateral). European SF CDOs tend to source larger portions of Consumer ABS, Prime RMBS, and CDOs2, as these sectors represent a significant portion of the European structured products market. The tables below provide a breakdown of recent collateral issuance for the US and European markets, as well as JPMorgan’s 2004 forecast as of year-end 2003 3. Table 2 and Table 3 provide volumes for the overall markets. Table 4 and Table 5 provide volumes for subordinate tranches and indicate the percentage of the total structure that is non-AAA. Table 2

US Structured Product Supply and Forecast ($Billion) 2001 % of $bn Total

RMBS: Prime RMBS: HEL Consumer ABS: Autos Consumer ABS: Credit Cards CMBS CDO Consumer ABS: Student Loans Other REIT Corp ABS: Equipment RMBS: MH

152.6 94.2 70.2 59.3 67.2 62.8 9.5 23.7 9.8 7.0 6.8

27.1% 16.7% 12.5% 10.5% 11.9% 11.1% 1.7% 4.2% 1.7% 1.2% 1.2%

2002

2003

$bn

% of Total

$bn

% of Total

228.9 159.0 88.2 65.8 52.1 58.5 19.5 6.6 10.6 5.9 4.6

32.7% 22.7% 12.6% 9.4% 7.4% 8.4% 2.8% 0.9% 1.5% 0.8% 0.7%

350.0 219.9 77.2 64.8 77.9 66.4 30.7 12.0 9.2 6.8 0.8

38.2% 24.0% 8.4% 7.1% 8.5% 7.3% 3.4% 1.3% 1.0% 0.7% 0.1%

2004 Forecast % of $bn Total

325.0 200.0 75.0 75.0 70.0 69.1 40.0 12.0 10.0 8.0 3.0

36.6% 22.5% 8.5% 8.5% 7.9% 7.8% 4.5% 1.4% 1.1% 0.9% 0.3%

Source: JPMS, IFR Markets, MCM, Bloomberg, Moody’s, Standard and Poor’s, Fitch.

Table 3

Europe Structured Product Supply and Forecast ($Billion)

RMBS: Euro (non-UK) CDO RMBS: Aussie RMBS: UK CMBS Corp ABS: Whole Business Consumer ABS: Student Loans Other Consumer ABS: Autos Other: Sov/Ag Consumer ABS: Credit Cards Corp ABS: Equipment

2001 % of $bn Total

27.7 29.0 9.5 8.8 14.3 7.8 0.0 12.0 4.4 11.8 3.3 3.0

21.0% 22.0% 7.2% 6.7% 10.8% 6.0% 0.0% 9.1% 3.3% 9.0% 2.5% 2.2%

2002

2003

$bn

% of Total

$bn

% of Total

34.6 22.2 11.0 11.6 13.7 10.7 0.5 10.4 8.9 9.9 5.8 3.3

24.2% 15.6% 7.7% 8.1% 9.6% 7.5% 0.4% 7.3% 6.2% 6.9% 4.1% 2.3%

83.8 28.8 21.5 19.6 15.5 14.2 4.5 12.3 4.7 9.8 5.9 3.2

37.5% 12.9% 9.6% 8.7% 6.9% 6.3% 2.0% 5.5% 2.1% 4.4% 2.6% 1.4%

2004 Forecast % of $bn Total

80.0 28.0 25.0 20.0 17.0 12.0 11.0 9.0 8.0 8.0 7.5 3.0

35.0% 12.3% 10.9% 8.8% 7.4% 5.3% 4.8% 3.9% 3.5% 3.5% 3.3% 1.3%

Source: JPMS, IFR Markets, MCM, Bloomberg, Moody’s, Standard and Poor’s, Fitch.

2. A significant portion of CDOs re-securitized in European SF CDOs are Small to Medium Entity (SME) CDOs, which are collateralized by receivables on a large number (typically thousands) of small business loans. The granular collateral pool makes arguably makes these CDOs more akin to ABS than to a traditional CDO.

6

3. Other includes Dealer Floorplan, Stranded Asset, RV, Boat, Consumer, EETC, Aircraft, Small Business Loan, Non-Performing, and Aircraft.

February 19, 2004 Analyst Christopher FlanaganAC [email protected]

Global Structured Finance Research CDO Research Structured Finance CDO Handbook

Table 4

US Subordinate (non-AAA) Structured Product Supply $bn

RMBS: HEL CMBS CDO Consumer ABS: Credit Cards REIT RMBS: Prime Consumer ABS: Autos Consumer ABS: Student Loans Other Corp ABS: Equipment RMBS: US MH

2001

9.5 13.0 12.5 10.0 9.8 5.0 2.4 0.3 6.9 1.3 1.4

% Sub

$bn

2002 % Sub

$bn

2003 % Sub

10.1% 19.3% 19.9% 16.8% 100.0% 3.3% 3.4% 2.8% 28.9% 17.9% 21.2%

18.4 9.6 12.4 10.5 10.6 6.4 3.2 0.6 2.0 0.6 1.1

11.6% 18.4% 21.3% 16.0% 100.0% 2.8% 3.7% 3.2% 30.8% 9.7% 23.5%

32.4 10.5 10.1 9.6 9.2 8.8 2.5 1.3 1.2 0.6 0.2

14.7% 13.5% 15.3% 14.8% 100.0% 2.5% 3.2% 4.3% 9.9% 9.1% 26.7%

Source: JPMS, IFR Markets, MCM, Bloomberg, Moody’s, Standard and Poor’s, Fitch.

Table 5

Europe Subordinate (non-AAA) Structured Product Supply $bn

Corp ABS: Whole Business CDO CMBS RMBS: Euro (non-UK) Other RMBS: UK RMBS: Aussie Other: Sov/Ag Consumer ABS: Credit Cards Consumer ABS: Autos Corp ABS: Equipment

5.0 5.8 7.3 2.5 4.9 1.4 0.7 5.6 0.1 0.2 0.2

2001

% Sub

$bn

2002 % Sub

$bn

2003 % Sub

63.6% 20.0% 51.4% 9.2% 40.4% 15.8% 7.8% 47.6% 3.5% 4.1% 8.0%

6.3 4.4 3.8 2.4 1.8 0.9 0.9 1.6 0.6 0.4 0.3

59.0% 20.0% 28.0% 7.0% 17.5% 7.9% 7.8% 16.6% 9.6% 5.0% 10.1%

6.1 5.8 5.6 5.6 2.7 2.7 1.9 1.4 1.2 0.2 0.1

43.1% 20.0% 36.3% 6.7% 22.2% 13.8% 8.9% 13.9% 19.8% 5.1% 4.3%

Source: JPMS, IFR Markets, MCM, Bloomberg, Moody’s, Standard and Poor’s, Fitch.

What's Not in SF CDOs and Why

Seasoned SF CDOs under-performed in 2003 due largely to exposures in esoteric ABS sectors such as manufactured housing, tobacco, aircraft, mutual fund fees, and franchise loans. With the possible exception of manufactured housing, exposure in esoteric ABS was generally limited to a small portion of the portfolio, but it was enough to impair performance in many deals. These small (and sometimes new) sectors were included in older vintage SF CDOs (2000-2002) because they offered some of the highest yields and added diversity to the portfolio. However, the drawback is that they were often unseasoned, with potentially flawed business models that take time to uncover. For example, franchise loan amounts were typically based on the business value, rather than the real estate value, leaving loans under-collateralized in the event of business failure. Another example is mutual fund fees, which were based on the assumption of a growing share of assets invested in equity funds, and did not anticipate the negative fund flows during the market downturn.

7

February 19, 2004 Analyst Christopher FlanaganAC [email protected]

Global Structured Finance Research CDO Research Structured Finance CDO Handbook

The manufactured housing sector accounts for the bulk of structured products tranches that have defaulted. The sector was severely hit with loose underwriting, irrational competition, overproduction, and high exposure to the weakest sectors of the economy in the past few years. The result was a “perfect storm” where external forces totally changed the rules of the industry causing performance to significantly diverge from original expectations. This was combined with deals that were structured with lower initial required loss coverage ratios from the rating agencies and lower excess spread requirements. Today’s “next-generation” SF CDOs contain less exposure to these esoteric sectors, and have instead substituted established sectors such as HELs and CMBS. Although trading is allowed in SF CDOs, all collateral concentrations are subject to limits/prohibitions set at the inception of the deal. Another exposure that caused problems in early SF CDOs was exposure to CDOs, many of which have performed poorly (especially US HY CBOs) over the last several years. As a result, most traditional US mezzanine SF CDOs have scaled back CDO collateral concentrations (e.g. from a 20% bucket to a 5-10% bucket). Others continue to allocate to this sector, reasoning that newer CDOs benefit from the “debtfriendly” structures that characterize today’s CDO market. European SF CDOs still often have moderate-sized CDO buckets, but these are typically concentrated in SME CDOs, which have developed a positive performance track-record. High grade SF CDOs may have 10-15% concentrations in non-PIKable AAA and AA tranches. We think that while a general reduction in CDO exposure is an understandable response to current investor sentiment, CDOs still make sense for the SF CDO structure, and there is no compelling reason to avoid CDO exposures once investor sentiment turns. Far Less Deterioration in Later-Vintage SF CDOs

Table 16 shows migration of the Weighted Average Rating Factor (WARF) versus the trigger WARF as set in the indenture for all SF CDOs rated by Moody’s between 1999 and 2002. The chart illustrates the problems in 1999 and 2000-vintage SF CDOs, which have rapidly deteriorating ratings on the underlying collateral. In contrast, 2001 and 2002-vintage SF CDOs have had far less rating deterioration.

Chart 15

Weighted Average Ratings Factor vs. Ratings Trigger (% compliance, negative number indicates non-compliance) 50 0 -50 -100 -150 -200

Whereas many esoteric ABS sectors -250May-02 Aug-02 Nov-02 Feb-03 May-03 Aug-03 Nov-03 are collateralized by business value 1999 2000 2001 2002 or depreciating assets, most sectors Source: Moody’s. in recent SF CDO transactions are securitized by real estate, which should provide a backstop on losses (the property value) in the event of default. Those assets not secured by real-estate (cards, autos, equipment) are some of the 8

February 19, 2004 Analyst Christopher FlanaganAC [email protected]

Global Structured Finance Research CDO Research Structured Finance CDO Handbook

most seasoned in the ABS market (although investors should still be diligent in evaluating the servicer risk in these sectors) and have well-established performance track records. SF CDOs Capitalize on Arbitrage Opportunities

SF CDO allocations to various structured product sectors change according to deal vintage, and deals are designed to capitalize on a market anomaly or arbitrage opportunity at issuance. To illustrate this point, we look at the HELs minus CMBS spread differential for the 2002-2003 period (Chart 16). Chart 17 illustrates the change in SF CDO collateral breakdown during this same period. Chart 16

BBB Spread Differential: HEL minus CMBS

Chart 17

Global SF CDO Collateral Distribution by Funded Volume: 2002 & 2003

bp 250

45% 40%

200

35% 30%

150

25% 20%

100

15% 10%

50

0 Jan-02 Apr-02 Jul-02 Oct-02 Jan-03 Apr-03 Jul-03 Oct-03

5% 0% RMBS

CMBS 2002

ABS

CDO

REIT

2003

Source: JPMS, IFR Markets, MCM, Bloomberg, Moody’s, Standard and Poor’s, Fitch.

During 2002, the exceptionally stable CMBS sector was pricing nearly in line with HELs, and SF CDOs concentrated 40% of their collateral in CMBS versus 18% in HELs. However, as spreads on HELs gapped out in 2003 (partially due to large issuance volumes), new SF CDOs altered their collateral makeup to exploit the opportunity, reducing CMBS exposure to 31% and increasing HEL exposure to 46%. As relative spreads and issuance continue to change in the structured products markets, we expect that new SF CDOs will adjust to achieve maximum advantage for investors. Of course, investors should perform appropriate due diligence to ensure that asset managers are staying within their area of expertise. Shifting Asset Allocation: A Real Life Example

Table 6 illustrates the substitution of HELs for CMBS as well as the decline in esoteric assets in four deals issued by Declaration Management and Research4 between 2000 and 2003. We selected this manager because they have been a repeat issuer in the market, with a transparent history dating back several years. Note the dramatic rise in HELs and decline in CMBS. Also note the significant decline in manufactured housing, aircraft, and other smaller structured product sectors.

4. Formerly known as Independence Fixed Income LLC. 9

February 19, 2004 Analyst Christopher FlanaganAC [email protected]

Global Structured Finance Research CDO Research Structured Finance CDO Handbook

Table 6

Investment Portfolios by Structured Product Sector Independence I Independence II (2000) (2001)

Current Avg Rating/Trigger Avg Rating RMBS: HEL RMBS: Prime CMBS RMBS: MH CBO (Baa) REIT Corporates Corp ABS: Structured Settlements Corp ABS: Aircraft Lease Corp ABS: Small Business Loan Other Consumer ABS: Auto Consumer ABS: Credit Card Corp ABS: Entertainment Corp ABS: Equipment Leasing Corp ABS: Franchise Loans RMBS: Property Tax Liens

Baa3/Baa2 18.92 10.63 20.14 13.31 4.92

12.29 3.62 4.15 2.41 4.66 0.05 0.81 3.85 0.24

Independence III Independence IV (2002) (2003)

Baa3/Baa2 Baa2/Baa2 % of Total Portfolio 22.44 25.09 11.56 13.14 31.35 32.29 14.09 11.61 3.95 4.01 1.71 1.65 0.63 7.50 0.61 3.47 2.19

1.32 2.48 1.55 2.95 2.00 1.10

0.50

0.37 0.44

Baa2/Baa2 55.48 13.73 9.88 5.28 4.97 4.75 2.50 1.25 1.16 0.99 0.01

Source: Fitch.

Collateral Quality: Weighted Average Rating

Since each of the aforementioned asset classes may issue notes of various ratings and with various levels of subordination, the quality of collateral is at least as important as the type of collateral. While traditional SF CDOs typically used collateral with weighted average ratings in the BBB vicinity (relatively higher spreads created attractive arbitrage), the advent of new securitization technology has allowed the development of “high grade” SF CDOs, which typically source AAA/AA collateral. Chart 18 and Chart 19 show weighted average ratings for cash and synthetic SF CDOs issued in 2003. Note that synthetics, with lower overall funding costs due to the presence of a large super senior tranche, are overwhelmingly high grade SF CDOs. Although cash deals remain largely backed by BBB collateral, the number of cash high grade deals has grown to 22%. This growth has been helped by the development of short term money market eligible tranches at the top of the capital structure. The advent of high grade SF CDOs bodes well for SF CDO supply, since there is a much larger supply of AAA/AA collateral, which typically makes up 80-85% of the capital structure in many structured product transactions. Supply of the (relatively smaller notional) subordinate tranches may be limited, especially in higher interest rate environments5. Potential lack of mezzanine structured products supply creates reinvestment risk in traditional managed mezzanine SF CDOs. Investors should insure that asset managers have sufficient access to collateral during both the rampup and reinvestment periods.

5. Although higher interest rates would also impact senior tranche supply, the reduction would have less impact on SF CDOs because supply is much larger in the first place. 10

February 19, 2004 Analyst Christopher FlanaganAC [email protected]

Global Structured Finance Research CDO Research Structured Finance CDO Handbook

Chart 18

Cash SF CDO Collateral Quality Breakdown: 16 Euro, 35 US deals ($29.1) BB 7%

B 5%

Chart 19

Synthetic SF CDO Collateral Quality Breakdown: 22 Euro, 4 US deals ($39.8) A 5%

AA 21%

A 14%

BBB 5%

AA 29% AAA 61%

BBB 53% Source: JPMS, IFR Markets, MCM, Bloomberg, Moody’s, Standard and Poor’s, Fitch.

Source: JPMS, IFR Markets, MCM, Bloomberg, Moody’s, Standard and Poor’s, Fitch.

11

February 19, 2004 Analyst Christopher FlanaganAC [email protected]

Global Structured Finance Research CDO Research Structured Finance CDO Handbook

SF CDO Assets: 101 Most structured products share certain similarities (ownership in a pool of receivables, subordination, sequential pay structure, overcollateralization), but there are also key differences across asset classes (specific assets, weighted average life, structural variations). The following pages provide an introduction to the elementary characteristics of structured products typically found in SF CDOs, including collateral, market participants, structure, and principal risks. This section may be skipped by investors that already possess a strong understanding of the structured products markets or used as a reference guide for those with a strong understanding of some sectors but not others. We cover four of the largest SF CDO collateral components (HEL, CMBS, US Prime RMBS, and UK RMBS) in some detail, and provide a brief overview of other structured products. To begin, Table 7 provides an overview of four key sectors. Table 7

Structure Comparison Major Currencies

CMBS

UK RMBS

US Prime RMBS

USD

USD

USD, GBP, EUR, CHF

USD

Typical New Issue Size $750mm

$1bn

$6-$7bn

$750mm - $1bn

Weighted Avg. life

3 - 5 years

5 - 10 years

1.0 - 5.5 years

7 years

Spread Indices

Libor, Swaps

Swaps

Libor, Euribor

Treasuries

Coupon Type

Fixed and Floating

Primarily Fixed

Fixed and Floating

Fixed and Floating

Rating Stability

Relatively more volatile due to sector consolidation and transformation

Very stable and Very stable and marked by upgrades marked by upgrades in seasoned in seasoned transactions transactions

Very stable and marked by upgrades in seasoned transactions

Typical Structure

AAA - 15.0% sub. AA - 10.0% sub. A - 6.0% sub. BBB 2.5% sub. (overcollateralization)

AAA - 16% sub. AA - 13% sub. A - 10% sub. BBB - 6% sub. BB - 3% sub.

AAA - 8% sub. AA - 5% sub. BBB 1.5% sub. (plus Reserve Fund)

AAA - 2.5% sub. AA - 1.25% sub.

Liquidity

Medium liquidity; largest issuers have the best liquidity

Very High

High; one of the most liquid sectors in Europe

Very High

Key Issuers

Ameriquest, Chase, Countrywide, GMAC/RFC, Option One

CSFB, Goldman, Lehman, Morgan Stanley

Abbey National, HBOS, Northern Rock

Chase, Countrywide, Washington Mutual, Wells Fargo

Source: JPMS.

12

HEL

A - .85% sub. BBB - .50% sub.

February 19, 2004 Analyst Christopher FlanaganAC [email protected]

Global Structured Finance Research CDO Research Structured Finance CDO Handbook

The largest SF CDO collateral sector is RMBS. Chart 20 illustrates the difference between the most common forms of prime and subprime RMBS to appear in SF CDOs. There are also regional differences. Below, we explore several RMBS sectors in detail. Chart 20

Types of RMBS

Borrower Lien LTV WAC FICO Originators

B&C Subprime ARM

2nd Lien

Prime Alt-A

Prime Jumbo

Credit

Prime

Documentation Property Type

Pristine Credit

1st

2nd

1st

1st

80%

90%

75%

75%

7.5%

9%-10%

6.5%

6.0%

630

715

715

725

Ameriquest Chase Countrywide

Chase Countrywide GMAC/RFC First Franklin

Countrywide GMAC/RFC IndyMac

Chase Countrywide GMAC/RFC Well Fargo

Impaired

GMAC/RFC Option One

Home Equity

Prime

Source: JPMS.

RMBS: Home Equity Loans (HEL) HELs are secured by residential real estate property, primarily first mortgages to borrowers unable to obtain prime funding due to the borrower’s credit (sometimes referred to as “B/C” loans). Second-lien loans are also included under the HEL moniker, although these are a much smaller proportion of the market. HELs are the largest sector in the public ABS market. Volumes have increased dramatically due to the current low interest rate environment and home price appreciation. Chart 21 below provides a brief overview of the HEL market.

13

February 19, 2004 Analyst Christopher FlanaganAC [email protected]

Global Structured Finance Research CDO Research Structured Finance CDO Handbook

Chart 21

Overview of the HEL Market

Certificates

Residential homeowners borrow funds which are collateralized by real property. Loan types include: Subprime B/C: •Fixed rate mortgage (FRMs) •Adjustable-rate mortgages (ARMs) 2nd Lien Mortgage: •Closed-end fixed rate •Revolving home equity lines of credit (HELOCs)

Lenders include banks and finance companies. To securitize, lenders sell a pool of receivables into a trust. They generally retain servicing of the loans. Lenders may alternatively sell the receivables to be packaged by Wall Street conduits. In this case, servicing responsibility is generally not retained by the lender.

Certificates are sold in the public and/or private markets and are secured by trust receivables. These certificates can have a variety of structures: • Fixed vs. floating coupon • Wrapped vs. senior/sub • Multi-tranche vs. single tranche Bond structure will depend on the underlying collateral, trust structure, and credit enhancement.

•Home improvement loans (HILs) •High LTV loans (125s) Source: JPMS.

Chart 20 provides an overview of different types of HELs. Note that Alt A and Prime Jumbos (addressed later on under the Prime RMBS heading) are included here for purposes of comparison. Most HELs are Floating Rate

14

HELs may be floating rate (Adjustable Rate Mortgage or ARM) or fixed rate (Fixed Rate Mortgage or FRM). ARMs currently account for approximately 75% of HEL volume. Most ARMs are indexed off 6-month LIBOR. Typically, ARMs are fixed for a period of time before resetting and are referred to as Hybrid ARMs. Types of Hybrid ARMs include 1/29, 2/28, 3/27, 4/26, or 5/25 loans, where the first number references the period the loan rate is fixed and the second references the period the loan rate is floating. Over the years, the majority of ARMs originated have shifted towards the Hybrid ARM product, with 2/28’s the most popular. After the initial fixed period, ARM rates are determined by adding the loan’s margin to the benchmark. On each reset date thereafter, typically every 6 months, the new rate is calculated and principal payments are adjusted so that the loan fully amortizes over its remaining life.

February 19, 2004 Analyst Christopher FlanaganAC [email protected]

Credit Enhancement

Global Structured Finance Research CDO Research Structured Finance CDO Handbook

Excess spread is the first line of defense against losses. It is composed of excess funds from interest paid on the HELs after expenses. Depending on deal structure, it may be (a) used to cover payment shortfalls in the current period, (b) used to maintain target overcollateralization levels, or (c) flow out of the structure to the residual class holder. Overcollateralization. Overcollateralization is equal to the par value of collateral minus the par value of issued securities. Overcollateralization may exist at inception or be built over time using excess spread. The overcollateralization “cushion” may be used to absorb principal losses as they occur. If overcollateralization levels are not in compliance with test levels, excess spread is used to accelerate principal payments to bondholders, thereby rebuilding overcollateralization levels. Subordination. More senior securities have principal/interest payments subordinated to their own, which provides a natural layer of protection because they are not impacted by loss that is absorbed by the layer or layers below them.

Prepayments/ Prepayment Penalties

HELs typically amortize over their term-to-maturity. Some loans require balloon payments (more common for 2nd lien) and others may have an interest only period. Loans may prepay in advance of the scheduled maturity due to voluntary prepayments (refinancing or sale of home) or involuntary prepayments (repossession or loss of home). When a prepayment occurs, principal is paid through to the security holders, thus retiring that portion of principal that is attributable to the loan that has prepaid. HELs exhibit much less interest rate sensitivity as compared to Conforming and Jumbo MBS. Contributing to the greater prepayment stability are: higher baseline speeds due to credit curing and the equity take-out component of the market, as well as prepayment penalties. Both FRMs and Hybrid ARMs have significant prepayment penalties. Penalties typically last 3-5 years for FRMs and the initial fixed period for Hybrid ARMs. An example of a typical prepayment penalty is 6 months interest on any prepayments in excess of 20% of the outstanding loan balance.

Net Interest Margin Securitizations

NIMs are typically securitizations of front-end residual cashflows (after bond coupons, fees, and losses) from a single unseasoned HEL deal. They are also entitled to overcollateralization releases after the step-down date as well as prepayment penalty income. NIMs typically achieve ratings of BBB by discounting the expected cash flow stream and applying various default, interest rate, and prepayment stresses. The notes have a weighted average life of 0.8 to 1.5 years and offer a spread premium of 250-300bp to like-rated ABS. The high spread premium is the primary reason that NIMs appear in CDOs. NIM performance has been excellent over the past few years due to declines in floating rate funding costs associated with an accommodative Fed policy.

15

February 19, 2004 Analyst Christopher FlanaganAC [email protected]

Global Structured Finance Research CDO Research Structured Finance CDO Handbook

RMBS: Prime6 Chart 22 below provides an overview of the Prime RMBS market. Chart 22

Overview of the Prime RMBS Market

Certificates

Residential homeowners borrow funds, collateralized by real property. Including: Conforming: Meet US Government Sponsored Enterprise (e.g. Fannie Mae, Freddie Mac) criteria. CDOs don’t typically have exposure to GSE loans.

Wall Street dealers purchase pools of loans and structure securities based on demand

Mortgage cashflows are carved up to address: •Prepayment Risks & Cash Flow Uncertainty •Call and/or Extension Risk •Negative Convexity •Specific client needs •Different Targets: Average Life, Yield, Duration, Credit Quality

Non-Conforming: Jumbo: Exceed the single-family loan limit for GSEs ($333,700 in 2004). Alt A: Issued to prime borrowers that have documentation or other non-standard loan characteristics. Source: JPMS.

Structure

The earliest, simplest of MBS classes, sequentials split pass-through cash flows into classes with different average lives. Sequentials are sensitive to prepayments, with all classes extending or shortening simultaneously. Floating-rate MBS may have a variety of average life profiles, some stable and some volatile. Coupons typically reset monthly based on a fixed spread over a specific index rate (often LIBOR). The maximum coupon (cap), spread over index, and average life profile are main determinants of yield. Floaters, due to their shorter duration, generally have less price volatility than fixed-rate MBS (unless rates rise and the coupon reaches its cap).

Prepayments

Borrowers have the right to prepay at any time without penalty — in effect “calling” their loans away from investors — prepayments may be partial or complete. Timing and rate of prepayments vary and produce non-level, less-predictable cash flows. Given current interest rate expectations and following rapid prepayments over the last few years, prepayments are generally expected to slow in the coming years.

RMBS: UK UK RMBS is one of the fastest growing sectors of the ABS market. UK RMBS is the largest sector in the European securitization market and a widely held asset class 6. A special thanks to Rajan Dabholkar and Eliza Hay for their help with the US Prime RMBS portion of this publication. 16

February 19, 2004 Analyst Christopher FlanaganAC [email protected]

Global Structured Finance Research CDO Research Structured Finance CDO Handbook

among European investors. It also provides an opportunity for investors to diversify away from US consumer credit into high quality secured asset-backed product. Chart 23 above provides an overview of the UK RMBS market. Chart 23

Overview of the UK RMBS Market

Certificates

The underlying collateral is prime UK mortgages for owner-occupied homes. Mortgage brokers are a vital source of new mortgage origination (up to 60% of new loans for some lenders), branches and telephone distribution are also important. UK Mortgage Product •Standard variable rate mortgages (capital repayment over 25 year life). •Flexible mortgages have experienced dramatic growth in the past few years. Allow borrowers to prepay their principal and feature “payment holidays” •“Fixed rate” or hybrid mortgages. Initial fixed rate reverts to floating after specified time period (2-5 yrs).

UK mortgage lenders can be divided into three basic categories: banks, building societies (operate under the principle of mutual ownership), and specialized mortgage lenders (a focus on “non-conforming” borrowers). The UK market is dominated by the largest lenders, which are primarily banks.

The master trust structure used in global UK RMBS creates a clean bullet repayment profile similar to that of credit cards. An issuer can issue multiple series of Notes from the master trust and principal payments from each series (as in credit cards) may be shared among series to create a bullet or meet a repayment schedule for an individual series.

Source: JPMS.

Despite the growth in new product types, UK mortgages typically have a number of common characteristics, including a term of 25 to 40 years, a floating interest rate, full amortization over the life of the loan, and a first charge (i.e., lien) on the property. The mortgage’s interest rate (the standard variable rate, or SVR) is set at the discretion of the lender and is loosely tied to the central bank rate. UK residential mortgages have historically been marked by stable mortgage prepayments. In recent years, the introduction of a number of flexible mortgage products and increased competition among mortgage lenders have led to increased remortgage activity among existing lenders. In turn, this remortgaging activity has pushed prepayment speeds marginally higher. Master Trust Structure

For investors familiar with the mechanics of a credit card ABS master trust structure, the UK RMBS master trust uses a similar technology to create a soft bullet repayment profile. Because the total amount of mortgage collateral in the trust exceeds the size of any RMBS tranche due, “prepayment leverage” ensures the trust can create a bullet payment in a short accumulation period. An issuer can issue additional series from the master trust by either adding new mortgage collateral to the trust or transferring part of the seller’s share to the investor share (since the seller’s share is very large when the master trust is first established). In anticipation of notes becoming due, principal payments are accumulated for the purpose of creating a bullet repayment for that series. 17

February 19, 2004 Analyst Christopher FlanaganAC [email protected]

Global Structured Finance Research CDO Research Structured Finance CDO Handbook

The seller’s share represents the total interest in the trust collateral retained by the seller and is set to a minimum amount. The seller’s share does not represent a form of credit enhancement, as it ranks pari passu with the investor share. Excess seller interest is the interest in the collateral retained by the seller that exceeds the required minimum seller interest. Excess seller interest absorbs temporary fluctuations in the collateral balance of the trust (i.e., higher than normal redraws). Collections on the mortgage pool are split between interest charges and principal repayments. Interest income is allocated on a pro rata basis among each series of the master trust. The allocation of principal collections depends on the cash flow stage of each series. Principal repayments may be used to 1) amortize pass-through notes or to accumulate principal for a bullet repayment, 2) pay the seller and reduce the size of the trust (provided the seller maintains a minimum seller’s interest, or 3) purchase new mortgage receivables from the seller during the “revolving period”. US$ Securities

UK RMBS typically issue a portion of their liabilities (typically shorter term) as US$ denominated, SEC registered securities. This is due in part to lower short term rates in the US, with more aggressive pricing in the US on securities inside three years. US$ tranches also help to diversify the UK RMBS investor base.

Credit Enhancement

UK RMBS use a senior/subordinate structure with credit enhancement provided by subordination and excess spread. In general, reserve funds provide liquidity and credit enhancement to the structure. The reserve fund provides liquidity in that it can be used to cover any interest shortfall on the notes and acts as credit enhancement in that it absorbs any losses in a calculation period to the extent that excess spread is insufficient. For master trusts with structured bullets, the reserve fund can also provide liquidity to pay AAA bullet principal (again, to the extent necessary). Additional structural features (eg, cash reserves and liquidity facilities) interact with the priority of payments to help ensure the timely payment of interest and principal to the notes.

RMBS: UK Non-Conforming UK non-conforming mortgages are analogous to US HELs. Like HELs, nonconforming RMBS are secured by residential real estate property, primarily first lien mortgages to borrowers unable to obtain funding from conventional mortgage lenders. Borrowers in the non-conforming market include self-employed (without sufficient proof of income and financial history), foreign nationals working in the UK (no credit record), and those with County Court Judgements (CCJs). In the UK, County Court Judgments are recorded when an individual has not repaid some form of debt, ranging from credit card bills to an unpaid mortgage. As of year-end 2003, the largest originators are GMAC RFC (RMAC), Kensington Group plc, and Britannia Building Society. RMBS: Continental Europe Together, German, Dutch, Spanish and Italian RMBS represent a large source of European structured products issuance. While mortgage characteristics (e.g., loan-tovalue, prepayments, fixed/floating, mortgage features, etc.) vary by country, the collateral is considered “prime” in that mortgage borrowers have not experienced past credit problems. The mortgages are typically first-lien (except Germany), 18

February 19, 2004 Analyst Christopher FlanaganAC [email protected]

Global Structured Finance Research CDO Research Structured Finance CDO Handbook

owner-occupied properties. Mortgage credit performance has been strong, and the sector has experienced a number of upgrades due to better-than-expected performance and prepayments (i.e., deleveraging of the structure). Highly rated European banks seeking diversified funding and regulatory capital relief have been the primary source of these deals. While structures also vary by country, continental RMBS transactions typically pay down senior tranches sequentially and may only amortize subordinate tranches provided that certain performance criteria are met.

CMBS7 The CMBS market emerged in the last decade as a response to the real estate cycle in the United States, as traditional lenders (commercial banks, insurance companies) avoided real estate exposure during the real estate downturn of the late 80’s and early 90’s. CMBS allows lenders to spread underwriting risk, diversify geographically, and increase liquidity. It has also increased funding options for developers and large owners of commercial property. Chart 24 below provides an overview of the CMBS market. Chart 24

Overview of the CMBS Market

Certificates

CMBS are backed by mortgages on commercial and multifamily properties that are incomeproducing and operated for economic profit. CMBS are backed by a wide range of property types (total portion of issuance in parentheses): • rental

apartments (22.4%)

• shopping centers and other retail facilities (36.2%) • office buildings (25.2%) • hotels (1.8%) • warehouse/industrial (7.9%) •nursing homes, mobile home parks and self-storage (6.5%)

Wall Street firms & other conduit operators securitize portfolios of newly originated loans to empty the warehouse and take profits Banks, thrifts and insurance companies securitize seasoned loans to clear the balance sheet, adjust exposures, or exit the sector Wall Street firms and real estate "opportunity" funds acquire and securitize portfolios of seasoned loans to finance the acquisition and/or cash out of the investment Owners of large commercial properties and pools of smaller commercial properties secure attractive financing as an alternative to a portfolio lender

Conduit/Fusion (67.9%): Conduit deals are well-diversified, about $1billion in total size. Fusion deals are conduit deals that include some large loans (>$50MM), which are typically high quality, shadow-rated investment grade loans but pose some concentration risk. Multi-borrower floater (19.3%): Backed by floating rate loans from several borrowers. They are generally shorter term deals with some adverse selection risk. Single asset/single borrower (9.8%): Either backed by a single property or a single borrower’s portfolio.

Source: JPMS.

Structure/Credit Enhancement

Prepayments/ Extension

Credit enhancement is achieved mainly through subordinated bond classes (AA to unrated classes). Default risk is a function of the initial LTV (Loan-to-Value) and DSCR (debt-service-coverage-ratio). Newly originated fixed rate loan pools carry significantly less prepayment risk, as there are large prepayment penalties to the individual borrowers, typically in the form of strong loan-level call protection and a lock-out period followed by defeasance. 7. A special thanks to Pat Corcoran and Yuriko Iwai for their help with the CMBS portion of this publication. 19

February 19, 2004 Analyst Christopher FlanaganAC [email protected]

Global Structured Finance Research CDO Research Structured Finance CDO Handbook

Extension risk is created when not enough income and equity are available to pay the balloon payment. CMBS with low LTVs have less extension risk. Regional Differences

European CMBS can be differentiated from US CMBS by several factors. First, US issuance is typically more standardized and uses the conduit structure. In Europe, different countries have different underlying lease terms and the Conduit market is smaller because banks have (historically) been able to fund real estate on balance sheet very cheaply. This is changing over time. Broadly, the European cash CMBS market can be broken up into the following categories: • Single borrower, single property (mostly UK based deals, with “trophy” assets). • Single borrower, single tenant with no disposal strategy on the properties (typically linked to rating of underlying corporate). • Single borrower, single tenant with a disposal strategy on the properties (i.e. sell the properties to pay the principal on the bonds). • Multi-borrower

CMBS: Interest Only Strips IOs are coupons stripped from an underlying pool of commercial mortgages. They allow an issuer to sell near par priced securities, even if the coupon on the underlying mortgages is greater than the bond coupons. A single IO strip (traditional form) is defined as the adjusted WAC of the loans minus the WAC of the principal bonds. An alternative form is two IO strips (PAC and Support). The PAC IO is stripped from the traditional IO. Its notional amount and size is determined assuming certain default and prepayment scenarios. PAC IOs generally have a WAL of 7.0 years. The Support IO is the leftover and bears most of the brunt in the event of early prepayment. CMBS IOs are typically limited to at most 5% of the SF CDO collateral pool. They are not included in diversity score or WARF calculation and are typically haircut for purposes of par value tests. CMBS: Non-Performing Loan (Europe) Following the introduction of the Italian Securitization Law in 1999, non-performing loans (NPLs) represented a large percentage of the Italian securitization market. A key driver behind NPL securitization was the favorable tax treatment that allowed Italian banks to amortize any losses (over a five year period) arising from sale or securitization of NPLs. However, this special tax provision was terminated in mid2001, which has caused Italian NPL securitization to decline. NPL securitizations are typically backed by commercial real estate properties. NPL securitizations frequently depend on a property disposal strategy for the repayment of bond principal and interest. Because the loans are non-performing, the loan servicer plays an important role in transaction performance (i.e., ensuring that the loans move quickly through any court proceedings). Going forward, German banks may begin securitizing non performing loan portfolios as well.

20

February 19, 2004 Analyst Christopher FlanaganAC [email protected]

Global Structured Finance Research CDO Research Structured Finance CDO Handbook

Real Estate Investment Trusts (REITs) REITs are companies that own and (in most cases) operate income-producing real estate, with assets in the $300 billion area. Common REIT property types include diversified pools of regional malls, shopping centers, office buildings, warehouses, and residential facilities. REITs are typically financed using about half debt and half equity (equity pays out at least 90% of its taxable income as dividends). This is a significant improvement versus the early 1990s, when individual properties were typically financed by mortgages with LTVs in the 90% area, leaving companies less flexibility and more exposed to interest rate risk. There have been no REIT bond defaults in the last 10 years. Positive performance can be attributed to conservative debt ratios, the ability to access the secured market in times of distress, and the property cycle upswing. SF CDOs invest in unsecured fixed rate REIT debt, the majority of which is rated BBB. In addition to an interest rate swap, the CDO structure typically also includes a timing swap to match semi-annual REIT payments to quarterly CDO payments. With BBB REITs typically pricing within +/- 20bp of CMBS BBBs (T+120 ten year average), the inclusion of REITs in a SF CDO is not yield enhancement. Rather, it helps to improve diversity, since rating agencies give some diversification credit versus CMBS.

Consumer ABS: Auto and Credit Card Auto ABS securitize secured consumer installment loans or leases used to finance new and used car purchases. Receivables carry a fixed interest rate and usually have a 36, 48, or 60 month term. Receivables are originated by captive manufacturer finance subsidiaries (GMAC, Ford, DaimlerChrysler, Honda, Toyota), banks (Chase, M&I, Regions, USAA), and specialty finance companies (AmeriCredit, Onyx, WFS). Auto ABS typically use the Owner Trust structure, which provides flexibility in structuring cashflows, permitting multiple senior tranches, as well as floating rate tranches. Credit enhancement to senior notes are typically provided by subordinated certificates supplemented with a reserve account. Loans may prepay in advance of the scheduled maturity due to voluntary prepayment (refinancing or sale of vehicle) or involuntary prepayment (repossession or loss of vehicle). Credit Card ABS are backed by receivables from unsecured consumer loans. Types of cards include revolving lines of credit (Visa and Master Card) as well as retail cards from issuers such as Macy’s and Neiman Marcus. Non-revolving charge cards from AMEX are also common. Large issuers include Chase, Capital One, Citibank, and MBNA. Nearly every Credit Card ABS issuer uses the master trust structure, similar to the UK RMBS structure discussed above. Structures include a revolving period of 1-10 years (where monthly principal collections are used to purchase new receivables), followed by an accumulation period. Notes may be fixed or floating, typically with a 5yr expected life. Excess spread, early amortization triggers, and subordination are key forms of credit enhancement.

21

February 19, 2004 Analyst Christopher FlanaganAC [email protected]

Global Structured Finance Research CDO Research Structured Finance CDO Handbook

Auto and Credit Card ABS are two of the largest, most mature, most liquid sectors in the ABS market. Both typically are priced at or near the tightest spreads in the ABS market. Because low spreads reduce CDO arbitrage opportunities, CDOs tend to limit exposure to most Auto and Credit Card ABS (although some is typical as a means of increasing diversity).

Consumer ABS: Consumer Loan (Europe) Credit card penetration among Europeans remains low compared to the US and amortizing term loans represent a large part of unsecured lending in many countries. Unsecured consumer loans may be used for an auto purchase, home improvement, or other reasons, although the loan’s purpose does not always need to be stated. Consumer loan ABS transactions frequently have a revolving period, during which prepayments are used to purchase new loan collateral. Cumulative loss rates for consumer loan ABS have remained very low (less than 2%), while excess spread has remained healthy (e.g., over 5%). Italian consumer loan performance varies according to the region where the loans are originated, with southern regions frequently experiencing higher default rates. Corporate ABS: Equipment Equipment ABS is backed by loan or lease receivables including agricultural, computer, industrial, medical, small ticket office, and trucking. Lessors include both independent leasing companies and captive subsidiaries of large manufacturing firms. Credit analysis includes a review of the projected remaining cashflows and underwriting standards. Leasing company receivables are typically diversified across geography, industry, and obligors, and additional analysis and credit enhancement are required in cases where concentrations are high. Lease payments typically cover over 90% of equipment costs and 75% of the useful life for financing leases (essentially a monthly payment plan). Operating lease payments cover less than 90% of equipment value. Residual value (estimated value at the end of the lease term) can be realized via a buyout option or sale, and may be given some credit in rating agency analysis. Credit enhancement includes subordination, reserve funds, and de-leveraging performance triggers. SF CDOs typically purchase the single-A tranche. Bonds are typically fixed-rate, with a spread pick-up of 6-8bp to like-rated autos, and have a weighted average life of two to three years and experience low prepayment volatility. Equipment ABS supply is typically correlated with performance of the overall economy, with businesses willing to add/replace equipment in robust economies. In Europe, large Italian lease companies have originated the majority of equipment lease ABS. To date, equipment, vehicles and real estate leases have backed these transactions.

Corporate ABS: Whole Business Securitization (Europe) Whole business securitizations are bonds that are backed by the cash flows from a company. WBS companies typically have very stable cash flows and usually benefit from regulation or other protections that make it unlikely that these cash flows will change. For example, water utilities, funeral homes, and pubs are all types of operating companies that have been securitized. Whole business securitizations 22

February 19, 2004 Analyst Christopher FlanaganAC [email protected]

Global Structured Finance Research CDO Research Structured Finance CDO Handbook

achieve higher ratings than unsecured debt through covenants (e.g. bondholders right to replace management) and committed liquidity. WBS may also include property or high profile assets that have real value for WBS bondholders. To date, whole business securitizations have been a UK phenomenon, due to the favorable legal status afforded to WBS bondholders.

CDOs Other CDOs may be included in the SF CDO structure. Unlike CDOs-of-CDOs, where nearly 100% of the collateral pool is other CDOs, SF CDOs typically source a limited amount of other CDOs as collateral. This number was often 20% in earlier deals, but has been closer to 5-10% in more recent deals. The purpose for including CDOs in the collateral pool may be either yield enhancement or diversification. In addition to overall CDO concentration limits, SF CDOs also have limits on PIKable collateral. CDOs of small- and medium-sized enterprise (SME) loans are common in Europe. SME CLOs are not arbitrage-driven, and the primary motivation for banks to do these deals is balance sheet relief. SME loan pools usually offer a very large number of underlying obligors and thus a high degree of granularity in the portfolio. Germany, the Netherlands, Spain and the UK have all contributed to SME CLO volume.

23

February 19, 2004 Analyst Christopher FlanaganAC [email protected]

Global Structured Finance Research CDO Research Structured Finance CDO Handbook

Does SF Collateral “Work” for CDOs? There is a good case, we think, for applying CDO technology to structured product collateral. Structured products offer a spread pick-up to like rated corporates, as well as comparable default and recovery rates. Low event risk, a relatively stable arbitrage opportunity, and low correlation with traditional CDO collateral are also positive.

Spread Pick-Up The rise in SF CDO issuance was primarily the result of a sharp widening in structured products spreads in the fall of 1998. At that point, there was a severe dislocation in many non-government markets that injected huge liquidity premiums into spreads and presented attractive arbitrage opportunities. The resulting exploitation of these opportunities resulted in the take-off of SF CDOs. Fortunately (for arbitrage purposes), structured products spreads have remained relatively wide, explaining their continued growth as an asset of choice in CDOs. We make the case later in this section that structured product performance has been in line with like rated corporates. As such, we believe the spread pick-up arises from relatively less liquidity and higher barriers to entry (complexity) in the structured products markets. SF CDOs, in essence, monetize this premium, which arises from several factors: • Small size of the subordinate structured product market8 • Small size of individual subordinate tranches • Small size of the subordinate structured product buyer base • Complex structures that require more sophisticated analysis To illustrate, Chart 25 shows a significant spread pick-up for BBB and single-A HELs to like rated finance corporates. By contrast, the most liquid structured products sector, Credit Cards, prices much closer to the corporate market (Chart 26). For this reason, SF CDOs typically have larger allocations to HELs and other less liquid sectors than to Cards. Chart 25

Chart 26

HEL vs. Finance Spread Differentials

Credit Card vs. Banks Spread Differentials

350

80

300

60

250 200

40

150

20

100

0

50 0

-20

-50

-40

-100

Source: JPMS.

Jul-03

Jul-02

Jan-03

Jul-01

Jan-02

Jul-00

Jan-01

Jul-99

Jan-00

Jan-99

Jul-98

Jul-03

Jul-02

Jan-03

Jan-02

Jul-01

Jul-00

Jan-01

Jul-99

Jan-00

Jul-98

Jan-99

Jan-98

HEL Minus Finance (A)

Jan-98

-60

-150

Credit Card minus Banks (A 5-year)

HEL Minus Finance (BBB)

Credit Card minus Banks (BBB 5-year) Source: JPMS.

8. Small size of both the subordinate market and individual subordinate tranches also (arguably) leads to scarcity, which has the impact of tightening spreads. This effect is in most cases outweighed by the liquidity premium. 24

February 19, 2004 Analyst Christopher FlanaganAC [email protected]

Global Structured Finance Research CDO Research Structured Finance CDO Handbook

Stability/Event Risk The CDO universe can be roughly divided into moderately leveraged sectors (HY CLOs at 12x, HY CBOs at 8x) and more highly leveraged sectors (investment grade corporate CDO at 25x, SF CDO at 20x). In general, highly leveraged sectors are able to apply greater leverage because they are backed by relatively more stable high grade assets. The problem with applying greater leverage to stable assets, however, is that these sectors become more exposed to single names and “tail” risk, where unusual (based on historical experience) scenarios can quickly eat through the small amount of equity subordination in the structure. Note that leverage can vary greatly across SF CDOs (8-25x), with less levered structures less exposed to event risk. The experience of IG CDOs in 2002 and 2003 — when incidents of fraud, defaults, and “fallen angel” 9 corporates were high — is illustrative. These unusual events (fraud in particular) were painful for IG CDOs. For example, in an IG CDO backed by 100 equally weighted corporates, a 4% equity tranche would be wiped out by four defaults (assuming zero recoveries). Although IG CDO portfolios may not have exposure to every case of fraud (e.g. Enron, WorldCom, Parmalat), exposure to even a few of these cases leaves little room for error. Although SF CDOs employ similar leverage to IG CDOs, they are less exposed to event risk. This can be largely attributed to the higher levels of granularity in the underlying structured products, which are often referenced to a large number of individual consumers. In addition, the underlying structured products are themselves credit enhanced to withstand multiples of base case scenarios. Some idiosyncratic risk is clearly present in the form of issuer concentrations (model risk) and servicer risk. These risks, however, can be mitigated by concentration limits, limiting exposure to established sectors with proven models, and investing in sectors with adequate back-up servicing capability.

Stable Arbitrage Opportunity The subordinate ABS and CMBS markets are particularly dependent on the CDO bid. We use the US HEL and CMBS market as an example, although a parallel argument could be made for Europe, where the CDO market has a similar importance. We’ve calculated that newly issued cash SF CDOs purchased approximately $10.2 billion in HEL subs and $7.9 billion in CMBS subs in 2003. In terms of 2003 issuance, this equates to approximately 32% of HEL subs and 80% of CMBS subs. The percentage of HEL subs was probably even higher in late 2003. Although CDOs purchase a significant portion of their collateral from the secondary market, these figures help put the CDO bid in context of the collateral market size. However, as we hope this paper makes clear, we believe the SF CDO market is here to stay, meaning the CDO bid is not expected to “go away.” SF CDOs Exert Technical Pressure On Collateral Markets

As such, the SF CDO bid can exert significant technical pressure on collateral spread levels. When collateral is cheap, CDOs will enter the market, supplying additional demand and effectively putting a cap on spreads. This effect is enhanced due to the lumpy (not smoothed over the year) nature of SF CDO supply, as large numbers of deals tend to begin ramping up collateral simultaneously when the arbitrage looks attractive. 9. Defined as a investment grade security (BBB and above) falling to high yield (below BBB) or default. 25

February 19, 2004 Analyst Christopher FlanaganAC [email protected]

Global Structured Finance Research CDO Research Structured Finance CDO Handbook

When collateral is rich, CDO demand decreases. Since the structured products markets are dependent on the CDO bid for subordinate paper, SF CDOs effectively put a floor on subordinate spreads at the point where the CDO arbitrage disappears. Normalized CDO Funding Gap 10

IG CDO

Dec-03

Jun-03

Sep-03

Mar-03

Dec-02

Jun-02

Sep-02

Mar-02

Dec-01

Jun-01

10 9 8 7 6 5 4 3 2 1 0 -1 -2 -3 Sep-01

Chart 27 shows arbitrage levels in terms of the CDO funding gap (by definition, spread on assets minus cost of debt) for both SF and IG CDOs. Because SF CDOs put “caps” and “floors” on collateral spreads, the arbitrage has been much less volatile in this sector over the last several years.

Chart 27

Mar-01

To a certain extent, this phenomenon can also be observed in the institutional leveraged loan market, where CDOs are significant buyers. In contrast, the CDO bid is less influential in investment grade CDS, due to the size of that market (approximately $5 trillion).

SF CDO

Source: JPMS.

Correlation Exposure to structured finance offers strong diversification benefits for CDO investors. This is because SF CDOs have exposure to several different asset classes (CMBS, Prime RMBS, HEL) that each have low correlation to traditional CDO asset classes (High Yield Bonds, Leveraged Loans, Investment Grade Corporates). Higher diversification reduces risks since it implies lower correlation in the variability of returns. Please note that correlation in this sense means spread correlation, and not default correlation. Table 8

US Monthly Spread Correlation (2000-2003): AAA/AA Structured Products & Traditional CDO Collateral

10 Yr AAA CMBS 10 Yr AA Prime RMBS 5 Yr AAA HEL JPM USD HY Index 5 Yr BB/BB- Lev Loan 10 Yr BBB Industrial

10 Yr AAA CMBS

10 Yr AA MBS

5 Yr AAA HEL

JPM USD HY Index

5 Yr BB/BBLev Loan

10 Yr BBB Industrial

1.0

0.5 1.0

0.2 0.3 1.0

0.4 0.4 0.1 1.0

0.4 0.4 0.3 0.1 1.0

0.4 0.1 -0.1 0.7 0.2 1.0

Source: JPMS, S&P LCD.

Table 9

US Monthly Spread Correlation (2000-2003): BBB Structured Products & Traditional CDO Collateral

10 Yr BBB CMBS 10 Yr BBB Prime RMBS 5 Yr BBB HEL JPM USD HY Index 5 Yr BB/BB- Lev Loan 10 Yr BBB Industrial

10 Yr AAA CMBS

10 Yr AA MBS

5 Yr BBB HEL

JPM USD HY Index

5 Yr BB/BBLev Loan

10 Yr BBB Industrial

1.0

0.4 1.0

0.4 0.1 1.0

0.3 0.4 -0.1 1.0

0.5 0.2 0.1 0.1 1.0

0.3 0.1 -0.1 0.7 0.2 1.0

Source: JPMS, S&P LCD.

26

10. Note: The actual funding gap values are normalized at 10. It is the yield on CDO collateral minus the cost of CDO liabilities, fees, and expected loss.

February 19, 2004 Analyst Christopher FlanaganAC [email protected]

Global Structured Finance Research CDO Research Structured Finance CDO Handbook

Table 10

Euro Monthly Spread Correlation (2000-2003): BBB Structured Products & Traditional CDO Collateral

Dutch 5-7 Yr BBB RMBS UK 5 Yr BBB Dollar RMBS Euro Leveraged Loan JPM EUR HY Index

Dutch 5-7 Yr BBB RMBS

UK 5 Yr BBB Dollar RMBS

Euro Leveraged Loan

JPM EUR HY Index

1.0

0.3 1.0

0.0 0.2 1.0

0.1 0.0 0.0 1.0

Source: JPMS, S&P LCD.

Table 11

Euro Monthly Spread Correlation (2000-2003): AAA Structured Products & Traditional CDO Collateral Dutch 5 Yr AAA RMBS Dutch 5 Yr AAA RMBS

UK 5 Yr AAA Sterling RMBS Pfandbriefe AA Financials

1.0

UK 5 Yr AAA Sterling RMBS 0.5

1.0

Pfandbriefe 0.1

0.3 1.0

AA Financials 0.2

0.1 0.4 1.0

Source: JPMS.

Default and Recovery Performance Any analysis of default and recovery in the structured product markets is challenged by a lack of historical data. Nevertheless, now that the structured products markets have matured and been tested through several economic cycles, we do have some (limited) history and experience to start examining defaults in the market and to assess the ultimate recovery on these bonds. Ultimately, default and recovery data allows investors to evaluate current SF CDO portfolio assumptions, and should help guide investment decisions, as defaults and recoveries play a key role in the structuring processes. Unlike corporates, where a default is caused by a discrete event like a bankruptcy or a missed coupon payment, structured products typically suffer principal losses over time as loans default and as those move through the cashflow waterfall. Thus, it may take months or even years before losses eat through the credit enhancement and bonds take their first dollar loss of principal. Even after the bond starts taking principal losses, it still may take months, or even to the end of the deal, before the ultimate loss of principal can be determined. In addition, missed interest and principal may sometimes be deferred to the next period or a later period. Missed payments may also sometimes be capitalized and repaid over the remaining life of the transaction. For our analysis of structured products defaults and recoveries, we use a recent Moody’s study of “material impairment” rates11, a concept that Moody’s has introduced to address the vagaries of default discussed in the paragraph above. One of the downsides of this study is that it addresses US structured products only, and nothing of its kind yet exists for the European market (largely due to the relative youth of the market). Nevertheless, we think the conclusions reached using the Moody’s US data are broadly applicable to European structured products given that 11. “Payment Defaults and Material Impairments of US Structured Finance Securities: 1993-2002”, Moody’s Investors Service, December 2003. 27

February 19, 2004 Analyst Christopher FlanaganAC [email protected]

Global Structured Finance Research CDO Research Structured Finance CDO Handbook

ratings are designed to have consistent meaning across regions. If anything, some conclusions are likely to be conservative given the exceptional rating stability of European structured products as shown in Appendix A. Material Impairment Rates

Materially impaired securities are defined as follows: • Securities that have not paid promised principal and interest in entirety by the final maturity date (unambiguously in default). • Securities that have sustained a payment default that has not been cured. • Securities that are rated Ca or C and hence are expected to suffer a significant level of payment losses in the future. Table 11 below shows the cumulative material impairment rates for both corporate and structured product securities (by sector). Table 12 shows the annualized material impairment rates by year, as well as the five year average. Table 11

Cumulative Material Impairment Rates of US SF & Corporate Securities by Original Rating, 1993 –- 2002 Year 1

Corporate Aaa 0.00% Aa 0.00% A 0.00% Baa 0.58% Ba 4.10%

Year 2

Year 3

Year 4

Year 5

Table 12

Annual Material Impairment Rates of US SF & Corporate Securities by Original Rating, 1993 – 2002 12 Year 1

Year 2

Year 3

Year 4

Year 5

5 Yr. Annual Average

0.00% 0.00% 0.00% 0.00% 0.00% 0.00% 0.00% 0.00% 0.00% 0.00% 0.21% 0.21% 1.20% 1.86% 2.78% 3.87% 8.67% 13.38% 16.97% 20.82%

0.00% 0.00% 0.00% 0.00% 0.00% 0.00% 0.58% 0.62% 4.10% 4.77%

0.00% 0.00% 0.00% 0.00% 0.00% 0.21% 0.67% 0.94% 5.16% 4.14%

0.00% 0.00% 0.00% 1.12% 4.64%

0.00% 0.00% 0.04% 0.79% 4.56%

ABS (including HEL) Aaa 0.00% 0.00% 0.05% 0.05% 0.05% Aa 1.29% 1.94% 2.27% 2.71% 2.83% A 0.05% 0.49% 0.87% 1.10% 1.10% Baa 0.25% 1.36% 2.30% 3.76% 4.38% Ba 1.41% 8.57% 14.38% 18.26% 19.70%

0.00% 0.00% 1.29% 0.66% 0.05% 0.44% 0.25% 1.11% 1.41% 7.26%

0.05% 0.00% 0.34% 0.45% 0.38% 0.23% 0.95% 1.49% 6.35% 4.53%

0.00% 0.12% 0.00% 0.64% 1.76%

0.01% 0.57% 0.22% 0.89% 4.26%

CMBS Aaa Aa A Baa Ba RMBS Aaa Aa A Baa Ba

0.00% 0.00% 0.00% 0.13% 0.29%

0.00% 0.00% 0.00% 0.41% 0.29%

0.00% 0.00% 0.00% 0.55% 1.47%

0.00% 0.00% 0.23% 0.83% 1.77%

0.00% 0.00% 0.23% 0.83% 1.77%

0.00% 0.00% 0.00% 0.13% 0.29%

0.00% 0.00% 0.00% 0.28% 0.00%

0.00% 0.00% 0.00% 0.14% 1.18%

0.00% 0.00% 0.23% 0.28% 0.30%

0.00% 0.00% 0.00% 0.00% 0.00%

0.00% 0.00% 0.05% 0.17% 0.36%

(prime mortgages) 0.00% 0.00% 0.00% 0.00% 0.00% 0.44% 0.00% 0.16% 0.47% 0.40% 1.33% 3.86% 0.00% 1.10% 3.32%

0.36% 0.89% 0.47% 5.06% 4.42%

0.58% 1.01% 0.63% 6.00% 4.98%

0.00% 0.00% 0.00% 0.40% 0.00%

0.00% 0.00% 0.16% 0.93% 1.10%

0.00% 0.44% 0.31% 2.56% 2.24%

0.36% 0.45% 0.00% 1.25% 1.14%

0.22% 0.12% 0.16% 0.99% 0.59%

0.12% 0.20% 0.13% 1.23% 1.01%

Source: JPMS, Moody’s.

Material Impairment Data is Conservative

Overall, the default history has been favorable, with structured products performing in-line with or better than the corporate market in most cases. Structural protections, including subordination and excess spread, have helped to insulate ABS investors from taking principal losses. However, we expect that some of these figures are conservative, for the following reasons. 12. Annual MI Rate for yr. n = 1- (yr. n survival rate / yr. n-1 survival rate).

28

February 19, 2004 Analyst Christopher FlanaganAC [email protected]

Global Structured Finance Research CDO Research Structured Finance CDO Handbook

A concentration of distressed sectors gives the “ABS” rate an upward bias. Table 13 shows that a large number of ABS Material Impairments come from sectors such as healthcare, franchise, and manufactured housing. Problems in these “story” sectors have been addressed in previous sections of this report, and, importantly, they are typically not included in today’s SF CDO collateral pools. At the risk of stating the obvious, material impairment rates would be significantly lower if these sectors were excluded from the calculation. Table 13

Number and Percentage of US ABS Material Impairments, 1993 - 2002 ABS Collateral Type

# Rated Securities in Sample

# Of Material Impairments

% Impaired

ABS Collateral Type

# Rated Securities in Sample

# Of Material Impairments

% Impaired

30

12

40.0%

Equipment

70

0

0.0%

Floor-plans Sml Business Loans Student Loans Other Receivables Other ABS

105

0

0.0%

108 343 68

0 0 0

0.0% 0.0% 0.0%

418

4

1.0%

Healthcare Receivables Franchise Loans Manufactured Housing Autos HEL

136

31

22.8%

661 623 2,320

80 9 31

12.1% 1.4% 1.3%

Leases Credit Cards

368 1,272

3 4

0.8% 0.3%

Source: Moody’s.

Structures and assumptions have been tested. As we have seen in our analysis, bonds have taken losses primarily when unforeseen market forces have severely impacted an industry (MH), rating agency assumptions about losses were inadequate (MH and HELs), and/or structures were ill equipped to handle losses (HELs). In addition, prime RMBS material impairment rates are inflated by losses in Quality Mortgage transactions, which are backed by assets that were arguably closer to subprime than prime at origination. While we cannot protect against unforeseen events, the market today is in better shape as the rating agencies have additional history to more accurately project losses and structures have been tested through several economic cycles. The market is now dominated by investment grade, well capitalized players. Many downgrades can be attributed to “problem issuers” that have since exited the business. HELs are a good example of a sector that has consolidated into stronger hands. Industry consolidation has weeded out the weakest players. Aggressive underwriting was a main cause of the downfall for many issuers (e.g. extending down to low quality borrowers, overstated appraisals, insufficient credit enhancement). Consolidation has left stronger, higher quality and well-capitalized players with more focused and disciplined business strategies (e.g., tighter underwriting, stepped up collections efforts). As a result, the industry is much healthier today with more resilient issuers to better handle the expected rise in losses in the context of weakening consumer credit. To whit, Chart 28 shows the top 10 home equity issuers in 1997 and 2003. Only three issuers were rated single-A or higher in 1997 vs. six in 2003. 29

February 19, 2004 Analyst Christopher FlanaganAC [email protected]

Global Structured Finance Research CDO Research Structured Finance CDO Handbook

Chart 28

Top 10 HEL Issuers in 1997 and 2003 1997 1997 Top Top 10 10 Issuers Issuers Issuer ContiMortgage Money Store IMC Advanta FirstPlus Conseco Finance UCFC GMAC Bank of America Amresco

2003 2003 Top Top 10 10 Issuers Issuers Rating Ba1 A1 NR Ba2 BBBaa1 Ba1 Aa1 A1 B1

$mm 5,685 5,505 4,857 3,150 2,983 2,734 2,725 2,685 2,674 2,291

% Market 9.0% 8.5% 7.5% 4.9% 4.6% 4.2% 4.2% 4.2% 4.1% 3.5%

Issuer GMAC Ameriquest New Century Countrywide Lehman Option One CSFB Washington Mutual JPMorgan Chase NovaStar

Rating A3 NR NR* A3 Aa3 BBB+ Aa3 A2 Aa3 NR*

$mm 30,714 23,302 17,369 15,721 14,482 11,008 6,692 6,674 6,482 5,694

% Market 15.6% 11.9% 8.8% 8.0% 7.4% 5.6% 3.4% 3.4% 3.3% 2.9%

*No debt outstanding

Source: JPMS, MCM.

All of the aforementioned factors point to the conclusion that the default rates implied by Moody’s material impairment study are probably conservative, and we don’t think it’s a far stretch to use slightly more aggressive (i.e. lower default rates) base-case assumptions for SF CDOs. However, we point out that the Moody’s data covers only a short credit cycle (albeit one that includes a severe downturn), and that it takes a few cycles to fully assess default risk. As such, a bit of conservatism is certainly healthy for CDO structures. Recovery Rates

Moody’s defines structured product loss severity as the present value of periodic losses (including both interest and principal shortfalls) as of the origination date. Defaulted securities accumulate their losses gradually, so complete information on recoveries is only available for a small sample of securities that have been paid down or permanently written-down (84 tranches). Recovery rates for these securities are shown in the “Zero Outstanding Balance” (i.e. actual final recovery rates) column in Table 14. Nevertheless, we can get a good idea of expected recoveries on a larger Table 15

Table 14

Projected Recovery Rate on HEL13

Recovery Rates by Sector, 1993-2002 Yr. After Origination Zero Out-standing > 6yr* > 8yr* Balance

ABS (including HEL) Mean Recovery 81.6% Standard Dev 27.4% # Observations 55 CMBS Mean Recovery 100.0% Standard Dev 0.08% # Observations 6 RMBS (Prime) Mean Recovery 73.0% Standard Dev 29.2% # Observations 136

66.1% 33.7% 24

59.4% 33.0% 20

100.0% 0.00% 4

100.0% 0.00% 4

67.5% 31.1% 102

54.5% 32.5% 60

Source: Moody’s, JPMS. * In addition to those with zero outstanding balance. includes impaired securities >n years past origination.

13. ABS Monitor, JPMS, 16 September 2003. 30

Recovery (%)

0%-10% 10%-20% 20%-30% 30%-40% 40%-50% 50%-60% 60%-70% 70%-80% 80%-90% 90%-100% 100% Average

# of Tranches Original Original Rtg. A Rtg. BBB

1 1 1 2 1 1 70%

Source: JPMS, Intex Solutions.

3 2 3 0 1 0 4 0 2 3 7 56%

February 19, 2004 Analyst Christopher FlanaganAC [email protected]

Global Structured Finance Research CDO Research Structured Finance CDO Handbook

sample by looking at “current loss-implied” recovery rates on impaired securities that are far away from their origination date (>6yr and >8yr column). Note that eventual recoveries will inevitably be lower than these “implied” numbers since losses will continue to accumulate over time. There is significant recovery rate volatility within each sector, as evidenced by standard deviations of 27% - 33% for ABS and RMBS. Table 15 looks explicitly at expected recoveries on HEL subordinates, and suggests similar conclusions about recovery volatility. Moody’s offers several explanations for recovery variation across deals. Among these is that securities with higher recoveries typically defaulted late, after much of the principal balance had been paid down. Securities with lower recoveries typically defaulted earlier or had quicker loss accumulation. For this reason, Moody’s notes that the recoveries on securities with zero outstanding balance are probably more severe than can be expected on a going forward basis, since this population represents the deals that were written down over a short period of time, and represents the more risky set of defaulters. Subordinate tranches are likely to have lower recoveries (less subordination and smaller overall size) than senior tranches. Table 15 supports this axiom for HELs. Because the Moody’s data set is largely representative of subordinate tranches (few seniors have defaulted), recovery rates may reasonably be adjusted upward for senior tranches. Putting It All Together: Annualized Loss Rates

Now that we’ve looked at both default and recovery rates for structured products, the question becomes what is a reasonable loss assumption for a SF CDO model. The answer, of course, depends on both the rating and the collateral. In Table 16 below, we work through an example for a hypothetical SF CDO with a Baa weighted average rating and collateral allocated as follows: 50% ABS, 30% CMBS, 20% RMBS. Note that we are using the annual material impairment rates from Table 12 and recovery rates from the zero balance column in Table 14. Table 16

Annualized Loss Rate for a Hypothetical SF CDO

ABS (including HEL) CMBS RMBS

Portfolio Weight

Annual Material Impairment Rate

Recovery Rate

Loss Rate

Aggregate Annual Loss Rate

50% 30% 20%

0.89% 0.17% 1.23%

59.4% 100.0% 54.5%

0.36% 0.00% 0.56%

0.29%

Source: JPMS, Moody’s.

We stress that the loss rate derived here stems directly from Moody’s studies. Undoubtedly, investors and SF CDO originators should make modifications to these assumptions. For example, at the risk of stating the obvious, the assumption of 100% recoveries for CMBS will likely prove unrealistic over a larger sample. In addition, we have already made a case that the annual material impairment rate may be overstated, particularly for ABS. These are but two examples, and other modifications are possible. Ultimately, our purpose for this study is to give investors and issuers the best information available regarding performance to date. This current “reality check” will prove particularly useful in evaluating the rating agency assumptions (see Rating Agencies section) used to structure SF CDOs. 31

February 19, 2004 Analyst Christopher FlanaganAC [email protected]

Global Structured Finance Research CDO Research Structured Finance CDO Handbook

SF CDO Structure To begin our analysis of SF CDO structures, Table 17 provides a comparison to traditional term ABS transactions, as well as CLOs (the most popular CDO sector). As seen, SF CDOs have longer average lives than traditional ABS, ranging from 6 to 14 years. While expected losses on the underlying collateral are expected to be fairly minimal, due to the leveraged nature of the transactions, triple-A credit enhancement levels are substantially higher than traditional structured products. Table 17

Structural Comparison: Traditional SF CDOs v Other CDOs & Structured Products SF CDOs

HY CLOs

Collateral

Mezzanine ABS (0 - 15 years)

Average Size

Credit Cards

HEL

BB/BB- Leveraged BBB / single-A Loans (5 years) credit default swaps (5 years)

Revolving unsecured credit

Term Loans (15 - 30 years) Secured by real estate

$400 mm

$400 mm

$1-4 bn

$775 mm

$750 mm

Cash flow structure

Amortizing

Amortizing

Bullet

Bullet

Amortizing

Average Life

6 - 12 years

6 - 12 years

5 - 8 years

3 - 10 years

0 - 10 years

Coupon

Primarily Floating Primarily Floating

Primarily Floating Primarily Floating Fixed and Floating

Credit Enhancement

Senior/Sub or Insured, Excess Spread, O/C

Senior/Sub or Insured, Excess Spread, O/C

Senior/Sub or Insured, Excess Spread, O/C

Senior/Sub, Senior/Sub or Spread Accounts, Insured, Excess Excess Spread Spread, O/C

Net Excess Spread

60 - 80bp

150 - 160bp

75 - 85bp

6% - 7%

3.5% - 4.5%*

Expected Loss 0.12% - 0.30%

0.45% - 1.00%

0.15% - 0.20%

5% - 8%

0.75% - 1.00%

Typical AAA Enhancement

20% - 27%

12%

10% - 20%

10% - 20%

20%

IG Syn CDO

Source: JPMS. * ARM HEL assuming flat LIBOR.

SF CDOs have higher subordination requirements compared to traditional structured products partly because the rating agencies use a conservative approach in evaluating them (see Rating Methodologies section). Furthermore, the tranches which make up the SF CDO portfolio themselves have structural and credit protection in the form of paid-in subordination, excess spread, overcollateralization, and cash flow diversion triggers, all of which are lacking in the collateral (e.g. consumer loans) used for traditional structured products. SF CDOs can be broken down into three different types: traditional cash, high grade cash, and synthetic, all of which have material differences, which are driven by the underlying collateral. Table 18 below compares the basic characteristics of each type in order to help investors better assess risk and value in each. Following Table 18, we provide an analysis of unique structural issues in structured products, and how these issues are addressed in SF CDOs.

32

February 19, 2004 Analyst Christopher FlanaganAC [email protected]

Global Structured Finance Research CDO Research Structured Finance CDO Handbook

Table 18

SF CDO Comparative Analysis Cash (Mezzanine)

Cash (High Grade)

Synthetic

Collateral &

Single-A/BBB tranches

AAA/AA tranches

AAA/AA tranches

Issuance 2003 ($bn)

44 deals (39 US, 5 Euro) $19.8bn

8 deals (7 US, 1 Euro) $8.4bn

25 deals (5 US, 20 Euro) $45bn (notional)

Slow/steady growth in issuance since early 2002.

Little issuance before 2003

Rapid growth, especially since mid 2002.

Transaction Size

$250 - $600mn

$800mn - $1.3bn

$600mn - $4bn

Cash Flow Structure

Amortizing

Amortizing

Bullet

Most common Purpose

Arbitrage.

Arbitrage or Balance Sheet

Balance Sheet

Typically managed with 10-20% annual substitution limit.

Typically managed

Typically static

Credit Enhancement (%)

AAA: AA: A: BBB:

AAA: 4-9% AA: 1-4% A: 0.5-2%

Super Sr: 3-15% AAA: 3-5%* AA: 1-4% A: 0.5-2%

15-21% 9-12% 8-10% 4-6%

*Effectively 'second senior'; less credit enhancement vs. cash. WAL

AAA: 4-6yr (1st AAA) 6-8yr (2nd AAA) AA: 7-10yr A: 9-10yr BBB: 9-10yr 3-5yr (w/ turbo-pay)

AAA: 1yr or less (MM) 6-8yr (2nd AAA) A: 7-10yr

Super Sr: AAA: 5-7yr AA: 5-7yr A: 5-7yr

Indicative O/C Triggers (min)

A/B* 103-105% C: 101-102%

MM and A-1: 103-105% A-2: 103-104% B: 100-102%

NA

*Some transactions carry a Class A O/C test (when the class B is PIKable), which is typically in the 106-110% area. Indicative I/C Triggers (min)

A/B: 113-116% C: 104-109%

Most do not require I/C tests.

NA

Liquidity

Less liquid than other more established CDO sectors (CLO/CBO), but improving.

New sector: little trading to-date. Relatively easier analysis of underlying HG structured products is positive for liquidity. Relatively smaller amount of term paper is negative for liquidity.

Synthetics issue very little in funded liabilities, hence the secondary market for synthetic paper may be limited. Generally less liquid than cash deals.

Source: JPMS.

33

February 19, 2004 Analyst Christopher FlanaganAC [email protected]

Synthetic SF CDOs: Credit Events

Global Structured Finance Research CDO Research Structured Finance CDO Handbook

Credit exposure in synthetic SF CDOs is governed by the credit event language in each transaction. In some ways, SF CDO credit events (i.e. events that result in the settlement of a credit derivative trade) mirror corporate credit event definitions as defined by the International Swaps and Derivatives Association (ISDA). However, there are important differences. We provide an overview below. For more information, see Moody’s Approach to Rating Synthetic Resecuritizations, October 29, 2003. Chart 29

Structural Overview of a Typical Synthetic SF CDO Portfolio credit default swaps

Portfolio credit default swaps Synthetic

BBB A

Tranched risk

Cheaper Liabilities

Super Senior

Typically Unfunded Swap

CDO AA AAA

Issuer Senior Mezzanine Equity

Funded Notes or Unfunded Swaps

Source: JPMS.

In most cases, SF CDOs use portfolio CDS (one contract for the entire reference portfolio) to create exposure. This structure is necessary because the single name market has struggled to develop since, although there are plenty of potential sellers (i.e. investors), there are few natural buyers of protection. In contrast, corporate synthetic CDOs typically use single-name CDS (one contract per reference entity) to create exposure. Unlike credit events in the corporate market, there is not yet an accepted standard (or set of standards) for structured products. Although the rating agencies have made some progress in defining appropriate structured product credit events and impose penalties (e.g. default or recovery rate stress) for non-compliance, the market is not yet uniform. The lack of an accepted market standard has two primary implications. First, investors must perform appropriate due diligence to ensure that they understand the credit event definitions in a particular transaction. Second, secondary market liquidity may be impaired to the extent that a transaction utilizes exotic definitions. Both the lack of consistency, as well as the relative youth of the structured product credit default swap (CDS) market, mean that there is a relatively smaller sample of actual triggered credit events, and less certainty as to how events will function in practice. However, due to the high credit quality (AAA/AA) of the reference entities and subsequent low probability of default, many investors place less emphasis on credit event definitions.

34

February 19, 2004 Analyst Christopher FlanaganAC [email protected]

Global Structured Finance Research CDO Research Structured Finance CDO Handbook

Another difference is that, unlike corporate credit events, which reference any obligation of the reference entity, structured product events reference a specific tranche within the structure. In other words, if there is a credit event in a mezzanine tranche and the SF CDO references a senior tranche, the credit event will not be triggered. This eliminates the “cheapest-to-deliver” option embedded in corporate CDS, which has been a factor in depressing recovery rates in that market. Below, we review credit event definitions for SF CDOs, which have been vital to the growth of this sector. Although definitions will vary from deal-to-deal, typical provisions are detailed below. • Failure to pay: This credit event applies to both interest and principal. Typical provisions exclude deferrable securities and temporary failures due to clerical or other errors. • Loss: This credit event is triggered by write-downs to the reference obligation, but is restricted to write-downs that are irreversible (i.e. excludes securities that include various provisions for investors to be made whole at a later date). This credit event may be settled in part (only marginal amount written down) or in full. • Rating Triggers: This credit event applies to securities that have been downgraded to Ca or below and do not experience either (a) loss or (b) failure to pay within a six month waiting period or their downgrade to a certain rating level (typically Caa2 or Ca). This provision addresses the back-ended losses that are exhibited in a significant number of distressed structured product tranches. Note that several typical/familiar credit events from the corporate market don’t neatly fit into the structured product definitions. For example, bankruptcy typically does not apply to structured products, which are designed to be bankruptcy remote. In addition, restructuring is also difficult to apply to structured products, since holders of the reference entity may have various non-credit related motivations for restructuring (including reputation, and incentives outside the structure) that can lead to moral hazard in decision to restructure. In some cases, bankruptcy and/or restructuring are included in SF CDO credit event definitions, primarily because regulatory bodies require institutions to have them for capital relief. In these cases, modifications are typically made to the event definition to mitigate the problems addressed above. As a final point, we note that credit events in a SF CDO can be settled in either physical (protection seller pays par and receives reference entity) or cash (protection seller pays par and receives current market value) form. The distinction is important since the reference entities may be illiquid. Market value risk (from the need to sell an illiquid security) is typically avoided as long as settlement is not forced before the earlier of legal maturity, acceleration or final workout. In other cases, rating agencies will typically haircut recovery rates to address market value risk.

35

February 19, 2004 Analyst Christopher FlanaganAC [email protected]

HG SF CDO: Short Term Tranches

Global Structured Finance Research CDO Research Structured Finance CDO Handbook

Investors that are uncomfortable with synthetic SF CDOs, but desire exposure to high grade structured products, may want to consider a high grade (HG) cash SF CDO. Like their synthetic counterparts, HG cash SF CDOs reduce their weighted average funding costs by employing a low-cost “super senior” tranche. However, they do so in entirely cash form by using short-duration money market or medium term note tranches. These tranches are senior to the term AAAs and pay sequentially. In limited cases (post reinvestment period, performing deal) they may pay pro rata for a period of time. Although not new to the CDO market, this type of funding has been appearing in several recent deals. Generally, money market tranches mature every three months to one year, leading to cheaper spreads than with term notes. Medium term notes typically mature every 2-3 years. This structure cheapens overall funding, but also introduces the risk that the short term paper, if re-issued at a discount, will eat into excess spread. This remarketing risk created by the short term tranches is typically assumed by a third-party liquidity provider, who must purchase the short-term notes if they cannot be successfully re-marketed below some maximum spread (ensures timely repayment of 100% of maturing notes). As such, the rating of the short term notes are tied to the ratings of the liquidity provider. Presumably, the high quality of the portfolio (AAA/AA) is a comfort to the liquidity provider, who holds downgrade risk (although it may be sold to a third party in some cases). We expect this is why most recent deals with money market features are backed by stable high quality collateral such as AAA/AA rated real estate ABS. Of course, it is possible to apply short term tranches to traditional subordinate collateral, although this would likely require incremental cost to the liquidity provider. The presence of a liquidity provider benefits the equity holder. Those holders would see excess spread deterioration without provisions for the liquidity provider to step in and prevent issuance at substantially wider spreads. Money market tranches are favored by bank (e.g. Citibank) liquidity providers that have ready access to liquidity at short notice. Medium term notes are favored by non-bank liquidity providers that don’t have commercial paper programs and have relatively less access to immediate liquidity. In a typical structure, at the time of writing, money market tranches are sold in the L minus 2 to plus 10bp area. The liquidity put costs an additional 20bp and a remarketing fees cost an additional 4-6bp, leading to an all-in cost in the L+22-36 area, well below spreads on a typical SF CDO senior tranche (L+60). If remarketing is unsuccessful, the tranche can be put to the liquidity provider in the L+40-70bp area (i.e. liquidity provider is a backstop buyer at an agreed upon price). We note that money market tranche price is also a function of size of issuance, with tighter pricing often achieved on larger tranches because, for example, many institutions are limited as to the percentage of a tranche that they are allowed to purchase.

36

February 19, 2004 Analyst Christopher FlanaganAC [email protected]

Global Structured Finance Research CDO Research Structured Finance CDO Handbook

In contrast, spreads on medium term note issuance are slightly wider (L+12), but the liquidity put is slightly less expensive (15bp) because the notes mature less often. As such, the total cost for medium terms notes is a few bps higher than for money market tranches. Traditional SF CDO: Senior/subordinate AAA structure

Senior and subordinate AAA tranches are becoming more common in traditional SF CDOs (unlike HG SF CDOs, both AAA tranches are term paper). In most cases, subordinate AAAs receive interest and principal after senior AAAs, and tend to price wider. These tranches have found demand from investors seeking to construct a barbell portfolio, who might use a subordinate AAA to overlay an existing position in lower quality CDO, ABS, or corporate paper. Such a fund benefits from a wider spread, but does not go down in rating. Senior AAAs typically have shorter maturities (4-6 years) and have seen demand from insurance and reinsurance companies, as well as other highly risk-sensitive AAA investors. One issue worth noting for senior AAA investors is prepayment risk, which is primarily a reinvestment concern from an investor perspective. In several recent deals, prepayment risk has been addressed by using a pro rata pay structure for all senior notes, subject to deal performance triggers being within compliance. This structure reduces reinvestment risk for senior AAA noteholders, and at the same time prevents the rising funding costs that would otherwise follow the amortization of the most senior (i.e. cheapest) tranche. As an added benefit, the resulting pro rata amortization of the single-A noteholders decreases credit risk for that tranche by reducing the notional balance.

Technical Note: “Second-seniors” are sometimes split-rated between AAA and AA based on differences in rating agency criteria. The critical difference between the rating agencies stems from the evaluation of loss. S&P rates a tranche by its likelihood of incurring “the first dollar of loss.” Put in the context of CDOs, the entire tranche above a certain attachment point can receive a AAA rating. Although based on a similar concept, Moody’s rates a tranche using the “expected loss” concept. At the heart of this rating methodology is the idea of loss as a proportion of the total investment. The result of the differing methodologies is illustrated in the hypothetical diagram below. Although the expected notional loss amount is the same for both the 20% and 10% “second-senior” tranches, it is larger as a proportion of the whole for the 10% tranche. For this reason, the 10% tranche may receive a AAA rating from S&P but a Aa1 rating from Moody’s. Split-Rated

All Triple-A

90%

Aaa / AAA

80%

Aaa / AAA

10%

Aa1 / AAA

20%

Aaa / AAA

37

February 19, 2004 Analyst Christopher FlanaganAC [email protected]

Global Structured Finance Research CDO Research Structured Finance CDO Handbook

In addition to customizing risk profiles for specific investors, in certain cases (such as with top-tier managers, or high quality portfolios), the average spread of a seniorsubordinate AAA structure may be slightly less than indicative AAA spreads, reducing the deal’s liability costs. This may be accomplished if the senior AAA is particularly large compared to the subordinate AAA, and prices especially tight. Therefore, although terms are deal specific, issuing senior-subordinate AAAs may allow certain deals some additional flexibility to purchase higher quality assets without impairing the arbitrage available to equity. In most cases, the aggregate amount of the AAAs is roughly equal to the size of a single AAA in a comparable deal. Since subordinate AAAs are generally not firstpriority tranches, their expected loss assumptions are generally higher than firstpriority AAAs, but of course still above the minimum required for a AAA rating. Extension Risk and the CDO Structure

Many structured products securities (which may, for example, have a 30 year legal final and five year average life) may extend beyond their expected maturity in the event that payment speeds or performance does not meet expectations. Any extension in the underlying structured products may result in an extension for SF CDO tranches, impacting the most subordinate tranche first, because CDOs pay principal sequentially. However, there are several factors that mitigate extension risk in both the structured product securities and the CDO structure: • Clean Up Call: Many structured products have a provision for a clean up call (at par value) at the option of the servicer or residual holder once the amount of collateral remaining falls below 10-20% of the original par amount. The callholder may exercise the clean up call to collapse the deal and avoid ongoing deal expenses. In addition, the call-holder has an incentive to exercise the call to preserve its reputation and insure that future issuance will price to call. • Prepayments: As discussed below under the interest rate cap heading, prepayment speeds in some structured products, particularly Hybrid ARMS (HELs), are quite fast due to refinancing activity, which mitigates extension risk. We do note, however, that the loans that do not prepay (left in the pool) may include those that who couldn’t refinance due to credit issues. These remaining loans typically have greater credit risk, which makes the clean-up call somewhat important. • CDO Structural Features: As discussed below, CDO auction calls exist to facilitate the liquidation of collateral once the CDO is past its expected maturity date. The mezzanine turbo feature uses excess spread to accelerate amortization of the most subordinate tranche, reducing extension risk. • Rating Agency Stress: SF CDOs typically assume that prepayments occur at the speed modeled in the transactions. The agencies then stress prepayment speeds. Moody’s applies a standard stress of up by 2x and down by half, and notes that more severe stress levels may be used for interest rate sensitive securities such as prime RMBS. Other structured products may have a bullet maturity (Cards, UK RMBS) or balloon payments with low prepayment assumptions (CMBS), which make extension less likely for these securities. Finally, all else equal, extension risk is more significant

38

February 19, 2004 Analyst Christopher FlanaganAC [email protected]

Global Structured Finance Research CDO Research Structured Finance CDO Handbook

for traditional mezzanine CDOs than for high grade CDOs, because mezzanine structured products tranches are more prone to extension. Auction Call

Most SF CDOs have a 35-40 year legal final, which is appropriate given the long maturity dates for some structured products. Despite their long legal-final, most SF CDOs have a maximum collateral weighted average life of 7-9 years from issuance, which makes the expected life on the CDO notes considerably shorter. The potential mismatch between the maturity of underlying SF CDO collateral (as much as 30 years) and the expected life of the CDO notes (7-10 years) is addressed explicitly in many SF CDOs. This is necessary because some institutions cannot buy notes with a long-dated legal final. To reduce the legal final, SF CDOs may include provisions for an auction call at the “expected” maturity (often 8 years). The auction call will be repeated semi-annually until it is “successful” (proceeds are sufficient to pay out all notes and accrued fees). To encourage timely execution, both senior and junior notes may include provisions for coupon step-ups each period that debt remains outstanding. Alternatively, equity coupon may be diverted either immediately or after a grace period following the expected maturity. The rating agencies do not give credit for the auction call in their ratings.

Mezzanine “Turbo-Pay”

Many deals include provisions that provide for early amortization of, or equity upside for, mezzanine note holders. These provisions are not uniform and may take different forms for high grade versus traditional cash SF CDOs. Traditional cash SF CDOs often place a cap on payments to equity holders (e.g. 10-20% per annum), which vary according to market clearing levels for the equity tranche. Excess returns above this amount are redirected to pay down or “turbopay” the most-subordinate rated note (generally BBBs, which are often a large percentage of the structure at 4-6%), which shortens their duration and effectively lowers the weighted average funding cost for the CDO going forward. The cap may continue until the stated tranche is paid down, or for a given number of years after issuance (paying down the next-most subordinate after the most-subordinate tranche is fully amortized). The equity cap is beneficial for mezzanine investors, and tranches with this feature typically exhibit a small spread give-up to like-rated longer duration securities. The turbo feature is generally acceptable to senior investors as well, because by retiring the most expensive notes, excess spread is increased, improving the funding gap (which is available to senior note-holders should collateral deterioration trigger a diversion of capital to pay-down senior notes). In addition, the presence of a turbo feature also is positive for senior noteholders because asset managers (especially those with significant equity holdings in the deal) will be less likely to “overheat” the deal by investing in risky collateral in search of high equity returns. Rating agencies may or may not give credit for a mezzanine turbo feature in their rating analysis. In cases where they do, it (somewhat paradoxically) makes the ratings more conservative. For example, Moody’s requires a mezzanine CDO 39

February 19, 2004 Analyst Christopher FlanaganAC [email protected]

Global Structured Finance Research CDO Research Structured Finance CDO Handbook

tranche to withstand certain loss based on its probability weighted average life, which is heavily weighted to the base case (turbo functioning, shorter tranches allow less expected cumulative loss). However, the stress that the tranche is subjected to is based off an average life that, in higher default scenarios, is much longer, because it does not benefit from the equity turbo (no excess spread available). The mezzanine tranches in High Grade SF CDOs are structured somewhat differently. Importantly, for many, the lowest rated tranche achieves a single-A rating and is a relatively smaller portion of the capital structure. These tranches often receive payments diverted from equity investors, but are more likely to conceptualize this payment as an option for “equity upside” than early amortization. These tranches typically receive some percentage of equity cashflows (rather than a cap feature) that can vary by deal. Diversity and Concentration Limits

SF CDO collateral tests are similar to other CDO sectors. Both O/C and I/C tests are present. O/C haircuts may apply to collateral rated below Baa3 (instead of below Caa1) in excess of some threshold. Most deals also have the standard tests for weighted average rating factor and name/industry/servicer concentrations. One significant structural modification for “next generation” SF CDOs is the reduction of single-issue concentrations to 1-2% from up to 5% in early deals. Also important, maximum servicer concentrations are down from 20% in some early deals to approximately 7% in newer transactions. This structural development has also occurred in investment grade synthetic CDOs, which suffered greatly from single-name exposures in the 2002 credit market downturn. Although we are strongly encouraged by the trend toward decreased single-name concentrations in investment grade synthetics (where the market has increasing depth), we are cautious of this trend for some SF CDOs. Our rationale is that the market is dominated by relatively few servicers and, at any given time, there may be a small number of issues available in the structured product market. As such, SF CDOs may be forced to either slow ramp up speeds (to address single-issue concentrations) or move into non-benchmark issuers and off-the-run names (to address servicer concentrations). In many cases, off-the-run names offer wider spread levels. To the extent that these levels are liquidity driven and not credit driven, an opportunity exists for CDOs due to their buy-and-hold focus. However, we stress that servicer concentrations should not be set so low as to force issuers/managers into poor credits or outside their area of expertise. A similar logic applies to the application of a minimum diversity levels across structured product “industries” (e.g. equipment, cards, CMBS, HEL)14. Older vintage SF CDOs attempted to diversify across industries (sometimes venturing outside of a managers area of expertise) to improve their diversity scores, which averaged about 25. Newer deals have accepted lower diversity scores, which are now in the 17-20 area.

14. See Rating Agencies section for a detailed description of the diversity score. 40

February 19, 2004 Analyst Christopher FlanaganAC [email protected]

HG SF CDOs: Building Overcollateralization

Available Funds Cap Risk and the SF CDO Structure

Global Structured Finance Research CDO Research Structured Finance CDO Handbook

High grade SF CDOs, like other CDOs, are subject to up-front fees (rating, underwriting, etc.) that reduce funds available to buy collateral at the inception of the deal. They also have a smaller equity component than many CDOs. As such, overcollateralization (which measures the amount of collateral as a percentage of rated debt) is reduced in HG SF CDOs because a greater proportion of the structure is rated. SF CDOs address potential undercollateralization with a provision to borrow under a swap agreement. The swap is then repaid at the top of the cashflow waterfall over time, reducing excess spread. Available funds caps are generally found in HEL ARM transactions, and limit the rate of interest that the deal is obligated to pay investors based on the interest accrued on the asset pool. The cap limits the investor coupon to the weighted average loan rate for the pool minus servicing fees and other costs. Caps on the underlying loans can be both “periodic” and “lifetime.” Periodic caps limit the amount the rate can reset upward at each reset date. Many of the loans have a higher periodic cap for the first adjustment period to allow the loans to more quickly adjust to market rates after being fixed for the initial teaser period. Life caps dictate how high the rates can reset over the life of the loan. Lifetime caps are generally quite high and therefore unlikely to be a limiting factor. On most deals a ‘shortfall reimbursement” feature covers any interest payment shortfalls by drawing on future excess spread. For the most part, fast prepayments have helped to mitigate most of the cap risk. Due in part to the teased nature of the product, ARM speeds have consistently come in above 50% CPR at the time of the first reset date. Because of the fast prepayments, by the time the caps start to matter later in the life of the transaction, the majority of the pool has already been paid off. In some respects, placing any significant value in the cap cost is a bet that ARMs will not continue to prepay quickly, which historically has not been the case in both the subprime as well the conforming ARM markets. Moreover, from our perspective, at the present time, there appears to be an irrational willingness to accept forward Libor rates as an interest rate forecast and at the same time assume that home price growth slows and employment growth remains weaken. While using forward Libor to value the available funds cap may be theoretically correct, it is nonsensical to assume that weak housing and labor conditions will prevail in a forward rate environment. Indeed, if that rate environment were to prevail, it would mean that the Fed is responding to exceptionally strong economic conditions, including home price and employment growth. Under such conditions, intense credit curing and equity takeouts would keep HEL prepayments at highly elevated rates and HEL loss rates at levels well below current rates. The market is missing this in the assumptions being made in its present valuations of HEL available funds caps. We draw two CDO related conclusions from the overvaluing of available funds caps:

41

February 19, 2004 Analyst Christopher FlanaganAC [email protected]

Global Structured Finance Research CDO Research Structured Finance CDO Handbook

Positive for CDO Equity: The overvaluation is most beneficial for equity, which receives an excess spread that has been driven higher by rising BBB HEL spreads, and represents the best arbitrage opportunity currently available. Low probability that CDO debt tranches will be impacted: There are several ways to mitigate exposure: 1) Hedge exposure by buying interest rate caps 2) The HELs with the most stringent caps have large fixed rate loans in the collateral pool. CDOs can reduce the cap issue by restricting HEL purchases to those with primarily floating rate pools. 3) Use a BBB turbo-pay feature to divert equity payments above a to the CDO BBBs, which would otherwise be the first rated tranche to miss coupon payments in the event that HEL coupon payments were insufficient to cover CDO liabilities. This strategy results in the early amortization of the BBBs and reduces the CDOs funding costs.

42

February 19, 2004 Analyst Christopher FlanaganAC [email protected]

Global Structured Finance Research CDO Research Structured Finance CDO Handbook

To conclude our section on structure, Table 19 below summarizes many of the differences between first and second generation SF CDOs that we have addressed throughout this paper. Table 19

SF CDO Comparative Analysis First Generation

Second Generation

Vintage Years

1999-2001

2002-2004

Funding Structure

Cash form

Various forms including traditional cash funding, cash with money market tranches, and synthetic.

Leverage

Typically 20x

Significant variation according to collateral quality

Collateral Sectors

Relatively higher exposure to esoteric structured product sectors such as MH, franchise, and mutual fund fees.

Most sectors are seasoned (less model risk). US deals have high percentage of real estate collateral.

Diversity

Often diversify across industries (sometimes venturing outside of a managers area of expertise) to improve their diversity scores, which averaged about 25.

Newer deals have accepted lower diversity scores, which are now in the 17-20 area.

Management

Manager ability impaired by poor liquidity in underlying collateral.

Collateral may be more liquid and better suited to a managed structure. Asset Managers have developed measurable track records and programmatic issuers have been identified.

Event Risk Structural Features

Significant servicer risk.

Less servicer risk in on-the-run sectors, which have more servicer depth. BBB-turbo, Auction Call

Source: JPMS.

43

February 19, 2004 Analyst Christopher FlanaganAC [email protected]

Global Structured Finance Research CDO Research Structured Finance CDO Handbook

Should SF CDOs be Managed or Static? SF CDOs may be managed or static. In 2003, approximately 75% of cash and 60% of synthetic transactions were managed. As a general rule, managed transactions are undertaken for the purposes of exploiting an arbitrage opportunity and generating fee income for the manager. In contrast, static transactions may be a funding vehicle or balance sheet management tool for the issuer. Managed Transactions

In managed deals, asset managers are typically allowed a 10-20% annual trading bucket as long as certain collateral quality tests are met (O/C, I/C, WARF, various concentration tests). Most deals have a reinvestment period of 3-5 years, after which discretional trading eliminated. Managers often have some flexibility (irregardless of reinvestment period or collateral quality tests) to sell defaulted and credit risk securities at any time. This flexibility is sometimes extended to credit improved securities. Senior management fees (paid at the top of the cashflow waterfall) are typically 1020bp. Subordinate management fees (paid at the after rated notes) are typically 2030bp. Both senior and subordinate fees have been falling for the last several years. In addition to holding a portion of the equity tranche, managers may also retain an incentive management fee equal to a percentage of excess cash flow after the equity tranche has received a benchmark return.

Static Transactions

Special Considerations

Although static transactions are not actively managed and have no reinvestment period, issuers may have the ability to substitute credit improved and/or distressed securities. In addition, issuers are responsible for the selection and/or origination of the original collateral pool. Static transactions feature reduced management fees. Unmanaged transactions have no reinvestment period, so senior noteholders may have shorter weighted average lives (and corresponding reinvestment risk) due to prepayments and natural amortization of the collateral balance. We feel compelled to note that there exists a potential for issuer moral hazard here, whereby the issuer includes less-desirable assets than they are willing to hold on their own balance sheet. As always, investor caution and appropriate due diligence is key. Ultimately, balance sheet deals are probably more appropriate for investors with some experience in the underlying assets that are able to carefully examine the pool without relying on an asset manager. Below, we explore several asset manager considerations specific to structured product collateral. Liquidity and Transparency. Structured products are typically less liquid/transparent relative to the corporate market. As such, managers have an increased opportunity to add alpha. In addition, there is significant price tiering among structured product issuers. Managers can also add alpha by selecting cheaper non-benchmark names. On the other hand, managers are somewhat restricted from active management due to the detrimental effect of crossing the (higher) bid/offer. Although the bid/offer spread is lower for senior structured

44

February 19, 2004 Analyst Christopher FlanaganAC [email protected]

Global Structured Finance Research CDO Research Structured Finance CDO Handbook

products tranches, the impact of crossing the bid/offer on excess spread is magnified because of the lower excess spread and higher leverage in these deals. Contagion Risk. Structured products are exposed to contagion risk, and securities with performance problems may exhibit “negative momentum”. Sectors with negative momentum benefit from active asset management, which can help minimize losses. Table 20 below shows 1-5 year cumulative material impairments for BBB structured products and corporates based on their original rating (at issuance) and cohort rating (at the start of the year, may be lower or higher than original rating). The table shows that material impairment rates for structured products are higher by cohort rating across all time horizons. The difference suggests that securities that have been downgraded to BBB from a higher rating (as shown by the cohort rating) fare worse than securities that were initially rated BBB. This momentum effect is not apparent in corporates. Table 20

Cumulative Material Impairments for All BBB Securities by Cohort and Original Rating Year 1

Year 2

Year 3

Year 4

Year 5

Structured Finance Cohort Rating Original Rating

0.99% 0.26%

2.53% 1.09%

5.01% 2.27%

6.49% 3.33%

8.36% 3.87%

Corporate Cohort Rating Original Rating

0.44% 0.58%

0.95% 1.20%

1.52% 1.86%

2.21% 2.78%

3.06% 3.87%

Source: Moody’s.

Delayed rating actions on structured product collateral. Rated noteholders depend on credit enhancement derived from the O/C tests, which divert excess cashflows in the event of collateral deterioration. Collateral, however, is held at par value or a haircut par value (e.g. sub-investment grade bucket) for the purpose of the O/C tests. This means that distressed collateral that is still investment grade will still be given “full credit” for the purposes of the O/C tests. This methodology is particularly problematic for structured products, where rating actions often severely lag the market developments. Asset managers can enhance rated-note stability via proactive management of the collateral pool before a downgrade occurs (realized losses are reflected in O/C tests). Programmatic Issuers. In both managed and unmanaged transactions, the selection of the issuer/manager is critical. Investors (especially those with limited structured products credit expertise) should favor high-quality “programmatic” issuers with a long-term commitment to the market. These issuers have an interest in preserving their reputation, as well as a visible track record, which minimizes any moral hazard in portfolio selection. In addition, programmatic issuers are familiar to the market, which increases liquidity of their notes. In some cases, because the SF CDO market is still relatively young, good managers of underlying structured

45

February 19, 2004 Analyst Christopher FlanaganAC [email protected]

Global Structured Finance Research CDO Research Structured Finance CDO Handbook

products are just now bringing their first SF CDOs. In these cases, it is especially important for investors to carefully evaluate the manager’s infrastructure and longterm commitment to the market. Infrastructure is particularly important due to the complexity of the collateral. Quantitative Manager Analysis

Table 21 below provides information on 37 managers of 58 SF CDOs issued between 1999 and 2002 for which Moody’s provides performance data. Although Moody’s data does not cover the entire issuer universe (e.g. it is only US), it does provide some of the more detailed performance measures available, these measures are briefly described below. It is important to note that these metrics measure performance on what we have referred to as “first generation” SF CDOs. Many lessons have been learned from the experiences. See Appendix C for a listing of seasoned SF CDO managers (managers with two or more deals in the last two years); this list includes some European managers. Moody’s Deal Score (MDS) MDS measures changes in expected loss for tranches that were originally rated investment grade. When current deal ratings deteriorate (expected loss increases), the MDS increases.15 Although ratings may lag performance, they can be a good indicator of manager ability over the long run. Annual Loss or Gain of O/C O/C is a measure of current collateral value relative to current liabilities. The O/C calculation measures most collateral at par value. Exceptions include defaulted securities (measured at lesser of market value or recovery rate). The table shows the annualized percent change in O/C. Negative numbers indicate that O/C is increasing. Weighted Average Rating Factor (WARF) Cushion WARF is an indicator of collateral pool risk. Higher rating factors indicate lower average collateral ratings. “Cushion” is the amount by which current levels exceed test levels imbedded in the deal. A negative cushion indicates that a deal is in violation of its test levels. Management flexibility is generally restricted following a breach of test level.

15. See Moody’s Publication CDO Rating Methodology: Moody’s Deal Score, February 2003, for more information. 46

February 19, 2004 Analyst Christopher FlanaganAC [email protected]

Global Structured Finance Research CDO Research Structured Finance CDO Handbook

Table 21

Moody’s SF CDO Asset Manager Analysis16 # of Deals

Current Moody's Deal Score (MDS)

Average Annual Loss/Gain of OC

Alliance Capital Management L.P.

1

0.00

-0.02%

5.62%

Asset Allocation & Management Company

1

2.29

-0.48%

-191.65%

Collateral Manager

Current Moody’s WARF Compl./Violation

Beacon Hill Asset Management, L.L.C.

2

0.91

0.20%

-102.43%

Capital Guardian Trust Co.

1

0.00

0.37%

-20.32%

Clinton Group, Inc.

2

0.00

-0.37%

-7.40%

Coast Asset Management, L.P.

1

0.00

0.37%

-17.21%

David L. Babson & Company Inc.

1

0.00

-0.04%

-8.49%

Deerfield Capital Management LLC

3

0.11

0.74%

-86.61%

Duke Funding Management , L.L.C.

1

0.00

-1.37%

-78.16%

Ellington Capital Management

2

0.00

-0.57%

-10.59% -37.75%

Fischer Frances Trees & Watts, Inc

1

0.00

-0.01%

Fortress Investment Corp.

1

0.00

0.23%

22.78%

General Re-New England Asset Management, Inc.

1

0.00

-0.40%

-29.34%

HarbourView Asset Management Corporation

1

0.00

0.79%

-109.87%

Hyperion Capital Management Inc.

1

0.32

0.52%

-79.85%

Independence Fixed Income Associates,

3

0.11

-0.86%

-54.83%

ING Baring (US) Capital Corporation

2

0.00

-0.11%

13.30%

Lord, Abbett & Co. LLC

1

0.00

-0.62%

2.49%

Metropolitan Life Insurance Company

1

0.00

0.89%

-18.66%

Metropolitan West Asset Management, LLC.

1

0.55

0.54%

-55.97%

MKP Capital Management, L.L.C.

2

0.00

-1.37%

-135.24%

MONY Life Insurance Company

2

0.00

-0.64%

28.31%

New York Life Investment Management LLC

1

0.00

0.40%

-16.25%

Newcastle Investment Corp.

1

0.00

-0.39%

23.25%

Pacific Investment Management Company

2

0.00

-0.11%

-3.18%

Phoenix Investment Counsel, Inc.

1

2.26

0.40%

-146.77%

PPM America Inc.

1

2.28

0.04%

-309.14%

Putnam Advisory Company, Inc.

2

0.00

-0.32%

9.82%

RWT Holdings, Inc

1

0.00

1.12%

7.74%

State Street Research

1

0.00

-1.14%

16.90%

Structured Finance Advisors, Inc.

2

1.48

0.68%

-67.76%

TCW Asset Management Company

4

0.00

-0.38%

-30.30%

Teachers Ins. and Annuity Assoc. of America

1

0.00

-0.05%

-83.59%

Wells Fargo Bank, N.A.

1

0.00

0.48%

-0.02%

West LB

4

0.06

0.17%

-5.01%

Western Asset Management Company

3

0.65

0.30%

-135.52%

ZAIS Group Inc.

1

0.00

1.60%

12.35%

Source: Moody’s Investors Service.

16. “Moody’s Deal Score Report”, Moody’s Investor Service, January 2004. 47

February 19, 2004 Analyst Christopher FlanaganAC [email protected]

Global Structured Finance Research CDO Research Structured Finance CDO Handbook

SF CDO Rating Methodologies Although assumptions vary by rating agency and by deal, to some extent SF CDOs benefit from conservative structures versus CDOs that source like-rated corporate collateral. Like all other CDOs, defaults, recovery rates, and correlation are key determinants in the credit enhancement process—below we provide a brief overview. Defaults and Recoveries

Due to the relative youth of the ABS market and the lack of a sufficient default history, Moody’s bases its expected loss assumptions on similarly-rated corporates. Moody’s does acknowledges that to date, ABS performance has been superior, and that their recovery rate assumptions may be “somewhat conservative” 17 (as follows, if you back out default rates holding expected loss constant, implied defaults are somewhat lower than corporates). Table 22 below illustrates Moody’s current recovery assumptions. Excessively thin tranches are penalized, as are CDOs with low Diversity Scores (reasonable, we think). Turning to S&P, less stringent default assumptions are used, but this is at least partially offset by more onerous ABS default correlation assumptions. Additionally, S&P uses a 7 year weighted average maturity for the purposes of determining subordination levels for SF CDO liabilities. This is particularly conservative for junior SF CDO Table 22

Moody’s Recovery Values for Structured Finance collateral Structured Finance Sector

% of deal

Tranche Rating

Diversified Securities1

> 70% 10 - 70% < = 10% >70% 10 - 70% 5 - 10% 2 - 5% < = 2% > 70% 10 - 70% 5 - 10% 2 - 5% < = 2% > 70% 10 - 70% 5 - 10% 2 - 5% < = 2% > 70% 10 - 70% 5 - 10% 2 - 5% < = 2%

Aaa 85% 75 70 85 75 65 55 45 85 75 65 55 45 80 70 60 50 30 85 75 65 55 45

Residential Securities2

Undiversified Securities3

Low Diversity CDOs4

High Diversity CDOs5

Aa 80% 70 65 80 70 55 45 35 80 70 55 45 35 75 60 50 40 25 80 70 55 45 35

A 70% 60 55 65 55 45 40 30 65 55 45 35 25 60 55 45 35 20 65 60 50 40 30

Source: Moody’s Investors Service. 1. Autos, Car Rental Receivables, Credit Cards, Student Loans. 2. Home Equity Loans, Manufactured Housing, Residential A/B/C. 3. CMBS Conduit, CTL, and Large Loan. Others not included in Diversified Securities. 4. CDOs with a Moody’s Diversity Score < = 20. 5. CDOs with a Moody’s Diversity Score >= 20.

17. “Moody’s Approach to Rating Multisector CDOs”, Moody’s Investors Service. 48

Baa 60% 50 45 55 45 40 35 25 55 45 35 30 20 50 45 35 30 15 55 50 40 35 25

Ba 50% 40 35 45 35 30 25 15 45 35 25 20 10 45 35 25 20 7 45 40 30 25 10

B 40% 30 25 30 25 20 15 10 30 25 15 10 5 30 25 15 10 4 30 25 20 10 5

February 19, 2004 Analyst Christopher FlanaganAC [email protected]

Global Structured Finance Research CDO Research Structured Finance CDO Handbook

tranches, which tend to have longer average lives than 7 years (say 10 years). These tranches are effectively penalized as they must withstand a higher scenario default rate than if their actual expected life were used to calculate subordination. Finally, Fitch also uses historical default rates of corporates as the basis for ABS— again given the performance of ABS (excluding CDOs) they reduce rates by as much as 20%. The agency also takes into account tranche size and seniority level to determine recovery rate—i.e. more senior (and/or larger) tranches would receive a higher rate. Correlation

Turning to default correlation, the rating agencies acknowledge that the lack of a default history poses a problem in accurately estimating joint-default probabilities of ABS sectors and issuers. S&P’s CDO Evaluator18 assumes a correlation of 30% for the same ABS sector within the same country, 20% for the same sector within the same region (i.e. UK RMBS vs Italian RMBS), and no correlation only in cases with the same sector but within different regions (i.e. US RMBS vs Australian RMBS). In contrast, S&P assumes no correlation between corporate sectors for other CDOs. For Moody’s, it’s more difficult to make an explicit comparison, but we describe their ‘Alternative Diversity Score’ methodology (adapted from the original framework for corporates). The calculation groups a portfolio into a matrix of ABS sectors and adjusts for par amount, default probability, and correlation. In the final analysis an actual, correlated portfolio is reduced to an idealized, uncorrelated one. Moody’s assumes lower joint-default probability for investment grade ABS than for sub-investment grade ABS and also makes specific adjustments—i.e. consumer sectors (such as Cards and Autos) might be highly correlated with each another, but as a group, largely uncorrelated with real estate sectors (such as CMBS and RMBS). Somewhat similar to S&P, Fitch uses a rule based approach that captures both regional and sector diversity—but all things equal, correlation between ABS is assumed to be higher than correlation between corporates. Correlation is assumed to be around 45% between ABS assets within the same sector and the same region. This assumption is reduced for assets from different sub sectors and/or different regions, and for CDO tranches the agency applies a ‘look through’ methodology that derives the correlation between CDO tranches, from the correlation between the underlying corporates. See Appendix B for an overview of rating agency classifications for structured securities.

Prepayment Stress

Finally, ABS collateral prepaying or extending at speeds different from what’s been originally modeled can be problematic, as it can impact the stability of the expected cash flows from the portfolio, affecting the credit quality of the CDO tranches. For example, a significantly higher prepayment rate than what was expected would lead to more asset cash flows than expected. If this is occurring in a declining interest rate environment, and/or during a slowdown in ABS issuance, the deal’s manager might have difficulty reinvesting this excess cash. As a result, Moody’s halves and

18. “Global Cash Flow and Synthetic CDO Criteria”, Standard and Poor’s. 49

February 19, 2004 Analyst Christopher FlanaganAC [email protected]

Global Structured Finance Research CDO Research Structured Finance CDO Handbook

then doubles prepayment speeds of ABS collateral. The other agencies perform something similar—for example S&P stresses the base case prepayment speed by 0.5 times and 1.5 times the base case to come up with slow (and fast) prepayment scenarios. Wearing our ABS hat, we think all of this is potentially more stressful than what has happened to ABS during different periods in the market’s history— but again highlight the benefit of this conservatism to CDO investors.

50

February 19, 2004 Analyst Christopher FlanaganAC [email protected]

Global Structured Finance Research CDO Research Structured Finance CDO Handbook

Appendix A: Rating Transition Matrices Table 23

Five-Year Transition Matrix for US Structured Products U.S. CMBS Five-Year Rating Transition Rate, 2003 and 1985-2003

AAA AA A BBB BB B

AAA 96.4 15.2 8.8 3.9 3.6 0.3

AA 3 77.9 14.1 11 1.6 1.4

A 0.2 3.4 70.1 9.2 2.8 1.6

BBB 0.4 0.9 3.8 66.7 9 4.9

BB

B

CCC/C

D

0.1 2.3 6.6 64.1 10.7

1 0.7 6.2 58.2

0.9 0.4 1.3 5.2 8.8

0.6 0.6 0.6 7.4 14

B

CCC/C

D

0.1 1.5 70.5

0.2 1.2 1.9 6.7 84

0.1 4.5 5.4

0.5 1.1 8 8.1 16

BB

B

CCC/C

D

0.1 0.4 2.1 42.3 15.4

0.5 2.9 2.9 49.2

0.7 0.6 2.3 3.8 6.2

0.5 1.5 4.4 8.2 12.2

U.S. ABS19 Five-Year Rating Transition Rate, 1982-2003

AAA AA A BBB BB B

AAA 99.5 8.7 3.3 6.2 2.7

AA 0.3 83.5 2.4 3.1 4.7

A 0.1 6.5 91.1 1.1 0.7

BBB 0.5 0.7 73.7 1.3

BB 0.1

U.S. RMBS20 Five-Year Rating Transition Rate, 1978-2003

AAA AA A BBB BB B

AAA 99.1 27.2 21 15.9 2.7 0.8

AA 0.6 66 14.2 13.4 10.8 0.9

A 0.1 4.7 59.3 12.6 11.4 3.6

BBB 0.1 0.8 2.5 46.4 17.8 11.6

Source: Standard and Poor’s.

19. US ABS includes: Mutual Fund Fees, Auto, Consumer ABS, Credit Card, Equipment, Franchise Loan, MH, Small Business Loans, Student Loans, Tobacco, Trade Receivables. 20. US RMBS includes: Prime (Jumbo/Alt A) and Subprime/Home Equity. 51

February 19, 2004 Analyst

Christopher Flanagan

Global Structured Finance Research CDO Research Structured Finance CDO Handbook

AC

[email protected]

Table 24

Lifetime Transition Matrix for European Structured Products CMBS

From AAA AA A BBB BB Ending Rtg. (#)

Starting AAA Rtg. (#) 94 100.0 68 5.9 86 4.7 79 1.3 41 2.4 368 104

AA

A

94.1 1.2 2.5

94.2 2.5

67

AA

BBB

BB

B

CCC

CC-D

83

93.7 2.4 75

90 37

2.4 1

0

2.4 1

A

BBB

BB

B

CCC

CC-D

Upgraded 0.0 5.9 5.8 6.3 4.9 16

Stable

Upgraded 0.0 26.3 45.2 29.0 85.7 39

Stable

Upgraded 0.0 3.6 11.6 3.1 0.0 30

Stable

Upgraded 0.0 0.0 0.0 0.0 0.0 0

Stable

Upgraded 0.0 0.0 4.8 4.5 0.0 9

Stable

100.0 94.1 94.2 93.7 90.2 350

Downgraded 0.0 0.0 0.0 0.0 4.9 2

Subprime RMBS

From AAA AA A BBB BB Ending Rtg. (#)

Starting AAA Rtg. (#) 59 100.0 19 26.3 42 23.8 31 7 158 74

73.7 21.4 3.2 24

54.8 25.8 14.3 32

71.0 71.4 27

14.3 1

AAA

AA

A

BBB

BB

89.0 3.6 3.7

5.5 87.9 7.9 1.9

5.5 8.6 87.2 1.2

1.2 96.9

287

156

174

159

100.0 17

2

AAA

AA

A

BBB

BB

B

95.7

4.3 100.0

100.0 73.7 54.8 71.0 14.3 119

Downgraded 0.0 0.0 0.0 0.0 0.0 0

Prime RMBS

From AAA AA A BBB BB Ending Rtg. (#)

Starting Rtg. (#) 310 140 164 162 16 765

B

CCC

CC-D

89.0 87.9 87.2 96.9 100.0 717

Downgraded 11.0 8.6 1.2 0.0 0.0 48

Corporate Securitizations

From AAA AA A BBB BB Ending Rtg. (#)

Starting Rtg. (#) 23 3 27 26 6 85

88.9

7.4 92.3

22

4

24

26

3.7 3.8 100.0 8

AAA

AA

A

BBB

BB

98.2 2.4 3.0

1.8 100.0 2.4 3.0

95.2 1.5

93.9

224

74

120

62

CCC

CC-D

3.8 1

95.7 100.0 88.9 92.3 100.0 79

Downgraded 4.3 0.0 11.1 7.7 0.0 6

Other ABS

From AAA AA A BBB BB Ending Rtg. (#)

Starting Rtg. (#) 223 67 125 66 5 488

Source: Standard and Poor’s.

52

1.5 100.0 6

B

2

CCC

CC-D

98.2 100.0 95.2 93.9 100.0 474

Downgraded 1.8 0.0 0.0 1.5 0.0 5

February 19, 2004 Analyst Christopher FlanaganAC [email protected]

Global Structured Finance Research CDO Research Structured Finance CDO Handbook

Appendix B: Rating Agency Classification of Structured Products Moody’s

Fitch

Standard and Poor’s

Asset Backed Securities Consumer finance-related instruments Auto (loan or lease) Credit Card Student Loan Home Equity Loans/Line of Credit Aircraft/Equipment Leasing Entertainment Royalties Small Business Loans Tax Liens Mutual Fund Fees

CMBS

Project Finance

RMBS Prime

CDO

RMBS Sub-Prime

ABS Consumer

Consumer ABS

ABS Commercial

Commercial ABS* Travel and Transport Small Business Others

CMBS Diversified (conduit and CTL) CMBS (large loan, single borrower, single property)

CDO of Corp.

REITS and REOCs

CDO of ABS

RMBS A

Structured Settlements Floor Plan Utility Stranded Cost Health Care Rental Car Consumer Mortgage-Backed Securities Conduit Large Loan Credit Tenant Lease Residential Mortgage Backed Securities Residential A Residential B&C

RMBS B&C, HELs, HELOCs, Tax Lien Manufactured Housing US Agency (explicitly guaranteed) Monoline/Financial Enhanced Rating (FER) Guaranteed Non-FER Company Guaranteed US FFELP Student Loans (over 70% FFELP)

REIT Debt Hotel Multifamily Office Retail Industrial Healthcare Self Storage Diversified Collateralized Debt Obligations Domestic Corporate Emerging Market Source: Moody’s, Fitch, S&P. * In Europe, no split between Commercial ABS sub-groups.

53

February 19, 2004 Analyst Christopher FlanaganAC [email protected]

Global Structured Finance Research CDO Research Structured Finance CDO Handbook

Appendix C: SF CDOs from Seasoned Issuers* Asset Manager/Issuer

Issue and Series

Year

Cash/ Synthetic

Region

American Capital Access American Capital Access American Capital Access American Capital Access Blackrock Financial Mgmt Blackrock Financial Mgmt Blackrock Financial Mgmt BNP Paribas BNP Paribas BNP Paribas BNP Paribas BNP Paribas BNP Paribas BNP Paribas BNP Paribas BNP Paribas BNP Paribas BNP Paribas BNP Paribas BNP Paribas BNP Paribas BNP Paribas BNP Paribas C-BASS C-BASS C-BASS C-BASS C-BASS C-BASS C-BASS Clinton Group Clinton Group Clinton Group Clinton Group Clinton Group Clinton Group Credit Agricole Indosuez Credit Agricole Indosuez Credit Agricole Indosuez Declaration Mgmt & Res Declaration Mgmt & Res Declaration Mgmt & Res Declaration Mgmt & Res Declaration Mgmt & Res Deerfield Capital Mgmt

ACA ABS 2002-1 ACA ABS 2003-1 ACA ABS 2003-2 Grenadier Funding I Anthracite CDO I Anthracite CDO II LEAFS CMBS Trust I CDO Master Investments II Iliad Investments Plc 1 Iliad Investments Plc 3 Iliad Investments Plc 4 Iliad Investments Plc 5 Iliad Investments Plc 6 Iliad Investments Plc 10 Iliad Investments Plc 11 Iliad Investments Plc 12 Iliad Investments Plc 14 Iliad Investments Plc 8 Iliad Investments Plc 9 Iliad Investments Plc 10 Iliad Investments Plc 11 Iliad Investments Plc 12 Iliad Investments Plc 14 C-BASS CDO I C-BASS CDO II C-BASS CDO III C-BASS CDO IV C-BASS CDO V C-BASS CDO VI C-BASS CDO VII Tribeca Mortgage Fund I 1 Bleecker Structured Asset Funding I Varick Structured Asset Fund, Ltd I Fulton Street CDO I Mulberry Street I Mulberry Street II ABSolute Synthetic CDO Ltd. I Triplas Synthetic CDO SA 1 Triplas Synthetic CDO SA 2 Seneca Funding I Independence CDO I Independence CDO II Independence CDO III Independence CDO IV Mid-Ocean CBO 1

2002 2003 2003 2003 2002 2002 2002 2001 2002 2002 2002 2002 2002 2003 2003 2003 2003 2003 2003 2003 2003 2003 2003 2001 2001 2002 2002 2002 2003 2003 1999 2000 2000 2002 2002 2003 2003 2003 2003 1999 2000 2001 2002 2003 2000

Cash Cash Cash Cash Cash Cash Cash Synthetic Synthetic Synthetic Synthetic Synthetic Synthetic Synthetic Synthetic Synthetic Synthetic Synthetic Synthetic Synthetic Synthetic Synthetic Synthetic Cash Cash Cash Cash Cash Cash Cash Cash Cash Cash Cash Cash Cash Synthetic Synthetic Synthetic Cash Cash Cash Cash Cash Cash

US US US US US US US Euro Euro Euro Euro Euro Euro Euro Euro Euro Euro Euro Euro Euro Euro Euro Euro US US US US US US US US US US US US US Euro Euro Euro US US US US US US

Source: JPMS, MCM, Thompson/IFR Markets, Moody’s, S&P, Fitch. *Includes all SF CDOs from issuers with two or more deals issued within the last two years. Due to lack of transparency in the CDO market, list may be incomplete for some managers.

54

February 19, 2004 Analyst Christopher FlanaganAC [email protected]

Global Structured Finance Research CDO Research Structured Finance CDO Handbook

Table (Continued) Asset Manager/Issuer

Issue and Series

Year

Cash/ Synthetic

Region

Deerfield Capital Mgmt Deerfield Capital Mgmt Deerfield Capital Mgmt Deutsche Bank Dresdner Bank E*TRADE AM E*TRADE AM Ellington Capital Mgmt Ellington Capital Mgmt Ellington Capital Mgmt Ellington Capital Mgmt Ellington Capital Mgmt Fisher Francis Trees & Watts Fisher Francis Trees & Watts Fisher Francis Trees & Watts Fortress Investment Group Fortress Investment Group Fortress Investment Group GMAC GMAC GMAC GMAC GMAC GMAC GMAC Gulf Intl Bank (UK) Ltd. Gulf Intl Bank (UK) Ltd. ING ING ING Lennar Partners Lennar Partners MFS IM MFS IM Pacific Investment Mgmt Co. Pacific Investment Mgmt Co. Pacific Investment Mgmt Co. Pacific Investment Mgmt Co. Putnam Investment Mgmt Putnam Investment Mgmt Putnam Investment Mgmt Putnam Investment Mgmt Redwood Trust Redwood Trust Royal Bank of Canada Royal Bank of Canada Royal Bank of Canada State Street Research

Mid-Ocean CBO II Oceanview CBO Ltd. I NorthLake CDO I Strips CDO 2002-1 Alexandria Capital 2003-1 E*Trade CDO 1 E*Trade CDO 2 Duke Funding, Ltd. I Duke Funding, Ltd. II Duke Funding, Ltd. III Duke Funding, Ltd. IV Duke Funding, Ltd. V Spinnaker I Commodore CDO I Commodore CDO II Fortress CBO Investments, Ltd I Newcastle CDO II Newcastle CDO III G-Force 2001-1 G-Force 2002-1 G-STAR 2002-1 G-STAR 2002-2 Blue Bell Funding I G-Force 2003-1 G-STAR 2003-3 FAB CBO BV 2002-1 FAB CBO BV 2003-1 Mane Funding I AJAX 1 AJAX 2 LNR 2002-1 LNR 2003-1 Crest Clarendon CRE CDO 2002-1 Crest Dartmouth Street 2003-I Pacific Coast CDO 2001 Euro Multi-Credit CDO I Pacific Shores I Pacific Bay I Putnam Structured Prod CDO 2001-1 Putnam Structured Prod CDO 2002-1 Putnam Structured Prod CDO (TAP) Putnam Structured Prods CDO 2003-1 Acacia CDO II Acacia CDO III Caribou 2002 Caribou 2002-D (Herald Ltd.) Caribou 2003-2 (Herald Ltd.) Fort Point CDO Ltd 1

2001 2002 2003 2002 2003 2002 2003 2000 2001 2002 2002 2003 1998 2002 2003 1999 2003 2003 2001 2002 2002 2002 2003 2003 2003 2002 2003 2002 2001 2002 2002 2003 2002 2003 2001 2002 2002 2003 2001 2002 2003 2003 2003 2003 2003 2003 2003 2002

Cash Cash Cash Cash Synthetic Cash Cash Cash Cash Cash Cash Cash Cash Cash Cash Cash Cash Cash Cash Cash Cash Cash Cash Cash Cash Cash Cash Synthetic Cash Cash Cash Cash Cash Cash Cash Cash Cash Cash Cash Cash Cash Cash Cash Cash Synthetic Synthetic Synthetic Cash

US US US US Euro US US US US US US US US US US US US US US US US US US US US Euro Euro Euro US US US US US US US Euro US US US US US US US US Euro Euro US US

Source: JPMS, MCM, Thompson/IFR Markets, Moody’s, S&P, Fitch. *Includes all SF CDOs from issuers with two or more deals issued within the last two years. Due to lack of transparency in the CDO market, list may be incomplete for some managers.

55

February 19, 2004 Analyst Christopher FlanaganAC [email protected]

Global Structured Finance Research CDO Research Structured Finance CDO Handbook

Table (Continued) Asset Manager/Issuer

Issue and Series

Year

Cash/ Synthetic

Region

State Street Research State Street Research Structured Credit Partners Structured Credit Partners Structured Credit Partners Structured Credit Partners Structured Credit Partners Structured Credit Partners Structured Credit Partners Structured Credit Partners TCW Advisors Inc. TCW Advisors Inc. TCW Advisors Inc. TCW Advisors Inc. TCW Advisors Inc. TCW Advisors Inc. TCW Funds Mgmt TCW Funds Mgmt TCW Funds Mgmt TCW Funds Mgmt TCW Funds Mgmt TIAA TIAA TIAA TIAA Trainer-Wortham Trainer-Wortham Trainer-Wortham Trainer-Wortham UBS Asset Mgmt UBS Asset Mgmt UBS Asset Mgmt UBS Principal Finance UBS Principal Finance UBS Warburg UBS Warburg UBS Warburg Vanderbilt Capital Vanderbilt Capital Vanderbilt Capital Wachovia Bank Wachovia Bank West LB West LB West LB West LB West LB West LB

Descartes CDO Ltd 1 Fort Point CDO Ltd II Crest 2000-1 INGRESS CBO, Ltd 1 Crest 2001-1 Crest G-Star 2001-1 Crest G-Star 2001-2 Crest 2002-IG Crest 2003-2 Crest 2003-I Eastman Hill Funding 1 South Coast Funding I South Coast Funding II Davis Square Funding, Ltd I South Coast Funding III South Coast Funding IV Westways Funding Ltd 1997-1 Westways Funding Ltd 3A Westways Funding Ltd 4A Westways Funding Ltd II Charles River CDO 1 TIAA Structured Finance CDO Ltd I TIAA Real Estate CDO 2002-I TIAA Real Estate CDO 2003-I TIAA Structured Finance CDO Ltd II Trainer Wortham 1st Republic CBO Trainer Wortham ABS CBO II Trainer Wortham ABS CBO III Trainer Wortham ABS CBO IV Diversified REIT Trust 1999-1 Diversified REIT Trust 2000-1 Brooklands Euro RLN 2001-1 Brooklands Euro RLN 2002-2 North St Ref Linked Notes Ltd 2003-5 North St Ref Linked Notes Ltd 2000-1 North St Ref Linked Notes Ltd 2000-2 North St Ref Linked Notes Ltd 2002-4 Bristol CDO I Grand Central CDO 2003 Lakeside CDO I Calibre 2003-1 Calibre II Blue Heron Funding I Blue Heron Funding II Blue Heron Funding III Blue Heron Funding IV Blue Heron Funding V Blue Heron Funding VI

2003 2003 2000 2000 2001 2001 2001 2002 2003 2003 2001 2001 2002 2003 2003 2003 1997 1998 1998 1998 2002 2000 2002 2003 2003 2000 2002 2003 2003 1999 2000 2001 2002 2003 2000 2000 2002 2002 2003 2003 2003 2003 2001 2002 2002 2002 2003 2003

Cash Cash Cash Cash Cash Cash Cash Cash Cash Cash Cash Cash Cash Cash Cash Cash Cash Cash Cash Cash Cash Cash Cash Cash Cash Cash Cash Cash Cash Cash Cash Synthetic Synthetic Synthetic Synthetic Synthetic Synthetic Cash Cash Cash Synthetic Synthetic Cash Cash Cash Cash Cash Cash

US US US US US US US US US US US US US US US US US US US US US US US US US US US US US US US Euro Euro US US US US US US US US US US US US US US US

Source: JPMS, MCM, Thompson/IFR Markets, Moody’s, S&P, Fitch. *Includes all SF CDOs from issuers with two or more deals issued within the last two years. Due to lack of transparency in the CDO market, list may be incomplete for some managers.

56

February 19, 2004 Analyst Christopher FlanaganAC [email protected]

Global Structured Finance Research CDO Research Structured Finance CDO Handbook

Table (Continued) Asset Manager/Issuer

Issue and Series

Year

Cash/ Synthetic

Region

West LB West LB Western Asset Management Western Asset Management Western Asset Management ZAIS Group LLC ZAIS Group LLC ZAIS Group LLC

Blue Heron Funding VII House of Europe Funding I Arroyo CDO I Pasadena (formerly Arroyo II) I Coronado CDO I Euro Zing I Euro Zing II High Tide CDO I

2003 2003 2001 2002 2003 2002 2003 2003

Cash Cash Cash Cash Cash Cash Cash Synthetic

US Euro US US US Euro Euro Euro

Source: JPMS, MCM, Thompson/IFR Markets, Moody’s, S&P, Fitch. *Includes all SF CDOs from issuers with two or more deals issued within the last two years. Due to lack of transparency in the CDO market, list may be incomplete for some managers.

57

February 19, 2004 Analyst Christopher FlanaganAC [email protected]

58

Global Structured Finance Research CDO Research Structured Finance CDO Handbook

February 19, 2004

Global Structured Finance Research 2004 Research Directory

G LOBAL S TRUCTURED F INANCE R ESEARCH Christopher Flanagan (1-212) 270-6515 Managing Director [email protected] Executive Assistant Carolina Sumaila (1-212) 270-5864 [email protected]

ABS/CDO RESEARCH Rishad Ahluwalia (London) (44-207) 777-1045 [email protected]

Edward Reardon (London) (44-207) 777-1260 [email protected]

Ryan Asato [email protected]

Parul Sahai [email protected]

(1-212) 270-0137

Benjamin Graves (1-212) 270-1972 [email protected]

Amy Sze, CFA [email protected]

(1-212) 270-0030

Ting Ko (London) [email protected]

Tracy Van Voorhis (1-212) 270-0157 [email protected]

(1-212) 270-0317

(44-207) 777-0363

CMBS RESEARCH Patrick Corcoran (1-212) 834-9388 [email protected] Yuriko Iwai (1-212) 834-9380 [email protected]

59

February 19, 2004 New York

Global Structured Finance Research CDO Research Structured Finance CDO Handbook

www.morganmarkets.com

Analyst Certification The analyst(s) denoted by an “AC” hereby certifies that: (1) all of the views expressed in this research accurately reflect his or her personal views about any and all of the subject securities or issuers; and (2) no part of any of the analysts compensation was, is, or will be directly or indirectly related to the specific recommendations or views expressed by the analyst(s) in this research.

J.P. MORGAN SECURITIES INC.

J.P. MORGAN SECURITIES LTD.

Tel. (1-212) 270-6000

Tel. (44 20) 7325 5080

270 Park Avenue New York, N.Y. 10017

125 London Wall London EC2Y 5AJ

J.P. MORGAN SECURITIES ASIA PTE. LTD.

J.P. MORGAN SECURITIES INC. (ASIA/PACIFIC) LTD. Chater House, 26/F 8 Connaught Road Central, Hong Kong Tel. (852) 2800-7000

Akasaka Park Building 2-20 Akasaka 5-chome Minato-ku, Tokyo 107-6151, Japan Tel. (813) 5573 -1185

Copyright 2004 J.P. Morgan Chase & Co. All rights reserved. JPMorgan is the marketing name for J.P. Morgan Chase & Co., and its subsidiaries and affiliates worldwide. J.P. Morgan Securities Inc. is a member of NYSE and SIPC. JPMorgan Chase Bank is a member of FDIC. J.P. Morgan Futures Inc., is a member of the NFA. J.P. Morgan Securities Ltd. (JPMSL), J.P. Morgan Europe Limited and J.P. Morgan plc are authorized by the FSA and JPMSL is a member of the LSE. J.P. Morgan Equities Limited is a member of the Johannesburg Securities Exchange and is regulated by the FSB. J.P. Morgan Securities (Asia Pacific) Limited (CE number AAJ321) is regulated by the Hong Kong Monetary Authority. J.P. Morgan Securities Singapore Private Limited is a member of Singapore Exchange Securities Trading Limited and is regulated by the Monetary Authority of Singapore (“MAS”). J.P. Morgan Securities Asia Private Limited is regulated by the MAS and the Financial Services Agency in Japan. J.P.Morgan Australia Limited (ABN 52 002 888 011/AFS Licence No: 238188) (JPMSAL) is a licensed securities dealer. Additional information is available upon request. Information herein is believed to be reliable but JPMorgan does not warrant its completeness or accuracy. Opinions and estimates constitute our judgment and are subject to change without notice. Past performance is not indicative of future results. The investments and strategies discussed here may not be suitable for all investors; if you have any doubts you should consult your investment advisor. The investments discussed may fluctuate in price or value. Changes in rates of exchange may have an adverse effect on the value of investments. This material is not intended as an offer or solicitation for the purchase or sale of any financial instrument. JPMorgan and/or its affiliates and employees may hold a position, may undertake or have already undertaken an own account transaction or act as market maker in the financial instruments of any issuer discussed herein or any related financial instruments, or act as underwriter, placement agent, advisor or lender to such issuer. Clients should contact analysts at and execute transactions through a JPMorgan entity in their home jurisdiction unless governing law permits otherwise. This report may have been edited or contributed to from time to time by affiliates of J.P. Morgan Securities (Far East) Ltd, Seoul branch. This report should not be distributed to others or replicated in any form without prior consent of JP Morgan. This report has been issued, in the U.K. only to persons of a kind described in Article 19 (5), 38, 47 and 49 of the Financial Services and Markets Act 2000 (Financial Promotion) Order 2001 (all such persons being referred to as “relevant persons”). This document must not be acted on or relied on by persons who are not relevant persons. Any investment or investment activity to which this document relates is only available to relevant persons and will be engaged in only with relevant persons. In other European Economic Area countries, the report has been issued to persons regarded as professional investors (or equivalent) in their home jurisdiction. Australia: This material is issued and distributed by JPMSAL in Australia to “wholesale clients” only. JPMSAL does not issue or distribute this material to “retail clients.” The recipient of this material must not distribute it to any third party or outside Australia without the prior written consent of JPMSAL. For the purposes of this paragraph the terms “wholesale client” and “retail client” have the meanings given to them in section 761G of the Corporations Act 2001. JPMorgan uses the following recommendation system: Overweight. Over the next six to twelve months, we expect this bond to outperform the average total return of the bonds in the analyst’s (or analyst’s team’s) coverage universe. Neutral. Over the next six to twelve months, we expect this bond to perform in line with the average total return of the bonds in the analyst’s (or analyst’s team’s) coverage universe. Underweight. Over the next six to twelve months, we expect this bond to underperform the average total return of the bonds in the analyst’s (or analyst’s team’s) coverage universe. JPMorgan uses the following rating system: Improving (I) The issuer’s long-term credit rating likely improves over the next six to twelve months. Stable (S) The issuer’s longterm credit rating likely remains the same over the next six to twelve months. Deteriorating (D) The issuer’s long-term credit rating likely falls over the next six to twelve months. Deteriorating+ (D+) The issuer’s long-term credit rating likely falls to junk over the next six to twelve months. Defaulting (F) There is some likelihood that the issuer defaults over the next six to twelve months. This report should not be distributed to others or replicated in any form without prior consent of JPMorgan.