CDO Handbook Final

FOR IMPORTANT DISCLOSURE INFORMATION relating to analyst certification, the Firm’s rating system, and potential

conflicts of interest regarding issuers that are the subject of this report, please refer to the Disclosure Appendix.

Cash Flow CDO HandbookStructures, Insights & Strategies

Structured Products

Editor’s Note In this publication, we have collected 25 reports on cash flow CDOs, most of

which were originally published in our bi-weekly “The CDO Strategist”. We have

also included some of our most popular previous publications as we believe

many of the concepts, ideas, risks, and structures are as relevant today. This

compilation covers almost all cash flow CDO types, ranging from Structured

Finance CDOs and CLOs to Trust Preferred CDOs; the topics include primers

on new products, investment strategies and relative value ideas, risks of CDO

products, and secondary valuation issues.

We hope our readers would find this handbook useful as a long-term reference

material and we believe some of the topics discussed will remain topical as the

CDO market evolves. We also welcome any feedbacks and comments so that

we can improve future editions.

Thank you.

31 March 2006

Fixed Income Research

http://www.credit-suisse.com/researchandanalytics

Contributors

David Yan

+1 212 325 5792

[email protected]

Stephen Chow

+1 212 538 5523

[email protected]

31 March 2006

Structures, Insights & Strategies 2

Introduction: 2005 Review, 2006 Forecast 3

Chapter 1. Structured Finance CDOs 18

Structured Finance CDO Primer 19

High Grade SF CDO Primer: Q&A 26

A Closer Look at High Grade SF CDOs 31

High Grade SF CDOs Revisited 42

Build or Buy: HEL Bonds versus SF CDOs 47

Revisiting Turbo Structure: Empirical Evidence 52

Auction Calls in SF CDOs 56

Default Assumptions for BBB HEQ in SF CDOs 65

Using the Right Rating Performance Measures of SF Securities for CDO Analysis 73

Impact of S&P’s New Rating Criteria on SF CDOs 84

Value Shifting to Mezzanine SF CDOs 92

Impact of HEQ Available Funds Caps on ABS CDO Tranches 96

Chapter 2. Collateralized Loan Obligations (CLOs) 110

Calling Attention to CDO Calls 111

When’s the Best Time to Call? Optimal Timing of CDO Calls and Relative Values 117

A Comparison of US and European CLOs 125

Chapter 3. Trust Preferred CDOs 137

Diversified Bank Trust Preferred CDOs - Primer 138

An Introduction to Insurance Trust Preferred CDOs 158

An Introduction to REIT Trust Preferred CDOs 176

Bank TruPS: Fine Tuning Historical Bank Failure Rates 190

Bank TruPS CDOs: Calling the Underlying 197

Chapter 4: Relative Value and Secondary CDO Market 201

Secondary Valuation Models of Cash Flow CDOs – Review and Pitfalls 202

2003 Vintage Mezz. SF CDOs – One of a Kind 212

Finding Value in Senior Tranches of Distressed SF CDOs 217

Seasoned Senior CLOs Should Trade Even Tighter 220

Junior AAA of HG SF CDOs Offers Attractive Value 223

31 March 2006


Introduction: 2005 Review, 2006 Forecast1

In 2005, the US CDO market drove home another record year despite some bumps on the

green. While uncertainty in the housing market mounted and the corporate credit

environment felt new pressures from big-name bankruptcies, sector-specific stresses, and

heightened leveraged buy-out (LBO) activity, the CDO market matured into a regular

fixture in the bond markets. Buoyed by innovation, investors searching for yield, and

robust demand, CDOs emerged not only as a balance sheet or arbitrage instrument, but

also as an efficient financing tool.

As we tee up for 2006, the CDO market faces off against new challenges in what’s

believed to be a more tumultuous course. What lurks around the corner? We take a look

back at 2005 and provide some thoughts on what lies ahead in 2006. We’ll share our “Top

5” issues for 2006 to help investors stay on the fairway.

CDO issuance: up, up, and away To call 2005 a record year for issuance would be an understatement. Not only did CDO

issuance surpass the record set in 2004, but the year-over-year growth was second only

to CMBS (which is partially attributed to the robust CDO demand for commercial real

estate [CRE] assets) among all other structured product and high yield (HY) primary

markets (Exhibit 1). The 2005 US CDO issuance volume reached a whopping $188 billion,

73% greater than 2004 by dollar amount and up 64% in terms of deal count (Exhibit 2).2

Exhibit 1: CDO issuance up 73% YOY, 2nd

highest growth among structured assets*

80%

51% 47%

31%23%

14% 11%

-38%-44%

73%

-60%

-40%

-20%

0%

20%

40%

60%

80%

100%

CMBS CDO MBS

(non-

agcy)

SL Auto CC HEQ Lev

Loans

HY

Bond

Other

ABS

*Year-over-year change by dollar amount issued; 2004 vs. 2005

Source: Credit Suisse, MCM, IFR, Bloomberg, Inside MBS & ABS, Fitch, Moody’s, S&P, BMA

1 This section was originally published in "The CDO Strategist", Issue #13, January 25, 2006.

2 Dollar amount includes MM/ABCP tranches and excludes unfunded tranches.

2005 issuance sets

new record: $188

BN across 368 deals

31 March 2006


Exhibit 2: CDO issuance reached a whopping $188 bn in 2005

$188 BN

$109 BN

(5.0)

5.0

15.0

25.0

35.0

45.0

55.0

65.0

75.0

85.0

95.0

105.0

115.0

125.0

135.0

145.0

155.0

165.0

175.0

185.0

195.0

1996 1997 1998 1999 2000 2001 2002 2003 2004 2005

Issu

an

ce (

$ b

illio

ns)

HY/EM CBO HY CLO IG CDO BAL SHEET SF/MS MV SYN OTHER

Source: Credit Suisse

On the cash side, CDO issuance remained dominated by structured finance (SF) CDOs

and HY CLOs. The “big two” combined accounted for $154 billion, or 82% of the CDO

market in 2005, nearly identical to 2004’s 83% market share (Exhibit 3). While the

absolute dollar increase was significant (70% growth over 2004), the relative market share

remained almost unchanged as several other asset classes made waves in 2005, which

we expect to ripple into 2006. In particular, the standardization of ISDA’s template for

CDS of ABS injected new fuel into an already energized synthetic CDO market, which

helped drive synthetic CDO issuance up considerably in ’05, as well as introduce the

hybrid cash/synthetic structure. Additionally, CDO technology was applied to new asset

classes such as REIT trust preferred securities (TruPS). We’ll discuss each asset class in

further detail below.

Exhibit 3: SF CDOs and HY CLOs continue to dominate issuance in 2005

15%11%

18%22% 22% 22%

26% 28% 29% 31%

12%

28%

42%46%

54% 51%

0%0%

0%

1%

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

1996 1997 1998 1999 2000 2001 2002 2003 2004 2005

% o

f Is

su

an

ce

$0

$20

$40

$60

$80

$100

$120

$140

$160

$180

Issu

an

ce ($

billio

ns)

SF/MS %

HY CLO %

SF/CLO $


SF CDOs – High Grade, CRE and CDS of ABS shape the market

For SF CDOs, several developments changed the landscape in 2005. We highlight three

key points as follows:

1. High Grade (HG) SF CDOs. HG SF CDOs experienced tremendous growth in

2005. Issuance totaled $51 billion, which represents more than 50% share of all

SF CDOs issued during the year. This also represents an 81% increase over

2004 by dollar amount and a 71% increase by deal count.

“Big two” issuance

soared, though

market share was

unchanged as other

CDO asset classes

showed prominence

in ‘05

31 March 2006


The economics of a HG SF CDO is largely dependent on how the largest tranche

– the senior tranche – is funded.3 For the past several years, term liability

spreads were too expensive to make term funding a widely used financing option.

Instead, HG SF CDOs turned to the ABCP market to fund the senior tranche.

This introduced additional transactional risks and costs, including 1) remarketing

risk (the money-market tranche must be remarketed every 270 days or so), 2) a

match-funding issue (use of short term paper to fund long-term notes), and 3)

counterparty risk from the put provider in place to purchase the notes should

remarketing be unsuccessful (the rating of the notes often shadows the rating of

the put counterparty). Counterparty risk was especially evident in 2005 as a

number of HG SF CDO tranches were downgraded because the put

counterparty’s ratings deteriorated.

As a result of spread tightening across asset classes, the funding paradigm has

shifted from ABCP to term funding as financing costs have converged. In 2005,

over 65% of HG SF CDOs were term funded (Exhibit 4). This mitigates the risks

associated with using ABCP as detailed above.

Exhibit 4: HG SF CDOs experience tremendous growth; shift towards term funding

14

11

1814

0

1

3

6

27

$1

$51

$28

$14

$4

0

5

10

15

20

25

30

35

40

45

2001 2002 2003 2004 2005

De

al C

ou

nt

$0

$10

$20

$30

$40

$50

$60 Iss

ua

nc

e ($

Billio

ns

)

Term Funded (deal ct - left axis)

ABCP Funded (deal ct - left axis)

Issuance ($BN - right axis)


Why was HG issuance so robust in 2005? We believe there are several reasons.

1) From a technical view, subordinate home equity (HEL) – which comprises

60%-90% of a typical mezz. SF CDO – spreads remained tight for most of 2005

and the arbitrage for mezzanine SF CDOs was squeezed, thus making these

transactions less attractive to equity investors; 2) Because CDOs were gobbling

up the overwhelming majority of subordinate HEL (upwards of 70%), collateral

sourcing became difficult; 3) From investors’ perspective, housing market

concerns may have pressured some investors to move up in credit, turning to HG

SF CDOs since liability spreads between the two products were trading on top of

each other – for most part of the year – while the spread volatility for senior HEL

tranches, compared to subordinate tranches, tends to be lower as they are more

cushioned in the event of a downturn in the housing market.

2. CRE CDOs. Another driver of growth for SF CDOs in 2005 was the impressive

upsurge in CRE CDO activity, which rose 84% to $16 billion (Exhibit 5). Several

factors contributed to this growth. First off, a large number of new entrants from

REITs to hedge funds entered the space. Participants realized the advent of

CDOs, not necessarily as a traditional arbitrage vehicle but rather as a low-cost

term financing vehicle without mark-to-market triggers as required by other

3 For a detailed discussion on HG SF CDOs, please see "The CDO Strategist - Issue #7 - A Closer Look at

High Grade SF CDOs", 9/20/2005, Credit Suisse CDO Research.

31 March 2006


funding alternatives. The collateral manager would retain the equity and/or junior

tranches, attaining attractive leverage while maintaining control of the loans in

case of work-outs should defaults occur.

Exhibit 5: CRE CDO issuance surge - $16 bn priced, an 84% rise

$1 $1 $3

$7$6

$9

$16

$0

$2

$4

$6

$8

$10

$12

$14

$16

$18

1999 2000 2001 2002 2003 2004 2005

Billions

0

5

10

15

20

25

30

35CRE CDO Issuance ($ BN, left axis)

CRE CDO Issuance (deal count, right axis)


Another contributor to the growth of CRE CDO was, quite simply, the sheer

increase in supply and demand. As we noted earlier, CDO YOY growth as a

whole was second only to CMBS in 2005. Although short-term interest rates

continued to rise throughout the year, long-term rates remained range-bound,

which supported an accommodative environment for CRE financing. Additionally,

continued strong performance of CMBS attracted new investors to the asset

class particularly from overseas and from CDO vehicles.

Furthermore, one particular development of CRE CDOs was the evolution of the collateral pool. CRE CDOs shifted away from traditional CMBS tranches and into non-rated CRE assets, including B-notes, mezzanine loans, and whole loans (Exhibit 6). There have even been transactions in ‘05 comprised almost entirely of whole loans, credit tenant leases (CTLs), or B-notes. As such, the demand from CDOs for unrated CRE assets has expanded that market beyond a handful of investors and into a much broader investor base.

Exhibit 6: Collateral pools shift away from CMBS and into un-rated CRE assets

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

2003 2004 2005

Non-CRE

Other CRE

Mezz

Whole Loan

B-note

REIT

CMBS

Source: Credit Suisse, Moody’s, S&P, Fitch, Intex

31 March 2006


3. CDS of ABS. No discussion of SF CDOs is complete without addressing the 100-ton gorilla that emerged in 2005: CDS of ABS. During the first half of 2005, ISDA published its template for executing CDS of ABS on a Pay-As-You-Go (PAYG) basis.4 This may well revolutionize the ABS markets and the impact has already been felt.

Hedge funds took full advantage of the development of the ABS CDS market, and CDOs were ready to respond. In October, macro hedge funds purchased protection on subordinate HEL tranches en masse, expressing a view of a possible downturn in the housing market. Subordinate HEL spreads in both cash and synthetic markets spiked to their highest levels since Q2 2004. BBB cash HEL spreads nearly doubled to 250 bps in the course of a couple of weeks.

In response, this brought about a renaissance for mezzanine SF CDOs, which had been relatively dormant due to collateral sourcing difficulties and the squeeze in equity arbitrage in the tight spread environment during most of the year. The emergence of Hybrid cash/synthetic structures presents investors another vehicle to take advantage of potential relative value opportunities between cash and synthetic assets. These structures have all the typical features of a cash CDO, including turbo features, OC/IC coverage tests, etc. with the ability to invest the majority of collateral synthetically, typically at a ratio of 70% synthetic and 30% cash assets. Moreover, several structures began featuring long/short strategies whereby the collateral manager could take opportunistic short positions to hedge single-name or market risk.

Issuance of cash and hybrid mezzanine SF CDOs doubled from Q3 to Q4 – from

$3.6 billion to $7.6 billion. Additionally, nearly $10 billion in cash or synthetic

mezzanine SF CDO transactions have been announced since the beginning of 2006.

The advent of CDS of ABS also greatly reduces the time required for deals to

come to market and the ramp-up risk associated with cash collateral sourcing.

Furthermore, CDO managers can now pick ABS credits selectively through the

virtually limitless synthetic supply and bring in more diversification, as opposed to

the limited cash market.

HY CLO – Thirsting for spread, weathering the storm, with 75 asset managers

The HY CLO market experienced another record-breaking year in 2005. Issuance

increased 82% by dollar amount to $58 billion across 112 transactions (Exhibit 7). Middle-

market loan (MML) CLOs also experienced relatively strong growth as the asset class

attracted a broader investor base.

However, the market was not without its bumps. Many credit events including big-name

bankruptcies such as Delphi, Delta Air Lines, and Refco, stresses in sectors such as auto

and aircraft, and intense LBO activity tested the resiliency of the CLO market. Despite

these events, CLO spreads remained firm or even tightened through the storm and only a

few transactions were impacted to the point of ratings downgrade. Several analyses we

conducted throughout the year showed that most CLO collateral pools were diversified

enough to withstand the credit events. Additionally, the benign economic environment

supported very favorable recovery rates, in the area of 80% for loans and 60% for senior

unsecured bonds, according to Moody’s.5

4 For more on the ISDA PAYG template, please see "The CDO Strategist - Issue #4 - An Introduction and

Comments on the New ISDA Template for CDS of ABS", 6/29/2005, Credit Suisse CDO Research. 5 Moody's "Monthly Default Report - November 2005", 12/6/2005. Figures represent 12-month trailing

recovery rates for US bonds/loans.

31 March 2006


Exhibit 7: CLO issuance grows 82%, in tandem with record loan issuance

$1 $4$13

$19 $17$13 $14 $18

$24

$46

$0$0 $1

$3

$7

$11

$0

$10

$20

$30

$40

$50

$60

$70

1996 1997 1998 1999 2000 2001 2002 2003 2004 2005

CL

O Issu

an

ce

($

Billio

ns)

$0

$20

$40

$60

$80

$100

$120

$140

$160

$180

$200

Institu

tion

al L

L Is

su

an

ce

($ B

illion

s)

MML CLO Issuance (left market)

HY CLO Issuance (left market)

Institutional LL Issuance (right axis)

Source: Credit Suisse, S&P LCD

Exhibit 8: CLO spreads tighten despite volatility in corporate credit

220

180

149

111

276

255

108100

125

150

175

200

225

250

275

300

325

9/3/04

10/3/04

11/3/04

12/3/04

1/3/05

2/3/05

3/3/05

4/3/05

5/3/05

6/3/05

7/3/05

8/3/05

9/3/05

10/3/05

11/3/05

12/3/05

1/3/06

Sp

rea

d (

bp

s)

BBB HY CLO

CDX IG 3-7%


Institutional leveraged loan spreads reflect the benign environment by remaining range

bound for most of the year and ending 2005 tighter on the year. The tight spread

environment posed a dilemma for both new-issue and secondary CLO transactions. In the

secondary market, vintage transactions in their reinvestment periods found it difficult to

meet the minimum weighted average spread (WAS) test. Managers dipped further down

in credit and looked towards riskier asset classes for spread within portfolio guidelines.

When this option was exhausted, many transactions opted to amend deal indentures to

lower the minimum WAS. Additionally, rather than sit on cash, equity holders of at least

27 CLOs past their non-call periods voted to call the deals in 2005, according to S&P.

In the primary market, deals were structured with much larger buckets for riskier asset

classes, such as mezzanine, second lien, and middle market loans, to enhance portfolio

spread. Whereas older deals (pre-2005) had buckets of around 5%-10% reserved for

such assets, deals in 2005 included buckets of 10%-20%. Particularly for middle market

and second lien loans, some transactions had buckets as high as 40%. Portfolio credit

also became riskier; Exhibit 9 shows the average WARF of HY CLOs in each vintage

versus BB institutional loans spreads.

31 March 2006


Exhibit 9: HY CLO WARF Increases: Portfolios get riskier, reaching for yield, as institutional loan spreads contract

2017

2103

2218 2214

2325

333

307

216202

185

1850

1900

1950

2000

2050

2100

2150

2200

2250

2300

2350

2001 2002 2003 2004 2005

Mo

od

y's

WA

RF

0

50

100

150

200

250

300

350

LL

Sp

read

(bp

s)

Avg CLO WARF (left axis)

BB Institutional Loan Spreads (right axis)

Loan spreads are from Credit Suisse Leveraged Loan Index, using year-end spreads for each year.

Source: Credit Suisse, Moody’s, S&P LCD, Intex

Further exasperating the spread conundrum was the fervent demand from an influx of

managers into the CLO space. In 2005, a whopping seventy-five managers crowded the

CLO market, with little, if any, spread tiering among new issue pricings. More so than ever,

manager selection becomes crucial, especially when managers are diving into riskier

assets. We’ll discuss this in further detail in our forecast section.

From a structuring perspective, 2005 saw an increase in the use of delay draw notes and

senior revolving structures, which benefits the equity by minimizing negative carry during

ramp-up. Additionally, market value structures and credit opportunity funds also saw

strong interest and as market value technology has expanded and evolved, it is being

applied to other CDO products as well. These transactions typically offer considerably

more manager flexibility and the ability to invest in more distressed pools at the expense of

lower leverage. Market value structures are in position to capitalize on market volatility,

which has increased in the corporate credit landscape in 2005 and is expected to rise over

the next few years as the turning of the credit cycle looms.

Finally, the seeds for leveraged loan CDS were planted in 2005 and synthetic exposure to

loans continues to gain momentum, absent a standardized template for these trades.

Almost all CLO transactions already have 10%-20% buckets available to take on synthetic

exposure to loans. With a standardized loan CDS template expected in the near term, we

believe this will be a robust market in 2006 with CLOs being the natural seller of protection.

TruPS CDOs: Innovation meets necessity

Trust preferred security (TruPS) CDOs experienced mixed issuance performance in 2005.

While total issuance across all TruPS asset classes rose to a record $9.7 billion, deal

count actually fell 11% to 17 deals from 2004’s high of 19. The reason behind this

discrepancy is two-fold. First, the average deal size of a typical TruPS CDO grew by 20%.

We believe this is because of better deal economics with larger pools and the fact that not

only were most TruPS CDOs comprised of hybrid pools of bank and insurance collateral,

but these transactions also began reaching into other asset classes because of

constrained collateral supply and tightening asset spreads. An increasing number of deals

began including TruPS CDO tranches and one transaction included a 12% concentration

in REIT TruPS.

Diversification,

collateral sourcing

difficulties, and tight

asset spreads drive

inclusion of other

asset classes in

TruPS CDOs

31 March 2006


This brings us to the second point: the application of TruPS on REITs.6 Similar to its

benefits for the banking and insurance sectors, TruPS CDOs provide relatively low-cost,

long-term, unsecured funding for REIT issuers, while providing investors the opportunity to

invest in pooled REIT risk with industry and geographic diversification at attractive spreads

and at new product premiums. Particularly for small to medium sized REITs (under $2

billion in market capitalization), TruPS help level the playing field by facilitating these

REITs with access to the capital markets in an efficient manner. In 2005, we saw four

REIT TruPS CDOs price, totaling $3.2 billion, a solid start for the asset class, although

collateral supply is constrained by the number of available REITs in the universe (currently

there are about 200 SEC-registered REITs and 800 unregistered REITs).

CDO credit: solid ratings performance, but sector & vintage matter In 2005, CDO downgrades continued to decline following a dramatic improvement in 2004,

driven by the benign economic environment, high recovery rates, and low long-term

interest rates (Exhibit 10). Downgrades averaged 29 tranches per month compared to 37

tranches per month during 2004 and 91 tranches per month in 2003. Deals on negative

watch, a precursor to future downgrade activity, also improved and stabilized last year.

On average, 29 deals were placed on negative watch each month in 2005 versus 37 deals

in 2004. Additionally, upgrades continued to edge up, with 307 tranches upgraded in 2005,

compared to 204 in 2004 (Exhibit 11).

However, sector and vintage clearly mattered as rating actions were not uniformly

distributed. SF CDOs accounted for over 67% of all downgrades in 2005 (by tranche

count), a considerable increase over the 46% share during 2004. Further examination

reveals that the surge is due to negative performance of troubled ABS sectors, such as

manufactured housing, among collateral pools in early vintage (pre-2003) SF CDOs. As a

result, 96% of downgrades among SF CDOs in 2005 occurred in these early vintages

(Exhibit 12).

Exhibit 10: Negative actions decline, but shift to SF Exhibit 11: Upgrades rise as deals de-lever

0

10

20

30

40

50

60

70

Dec-03

Feb-04

Apr-04

Jun-04

Aug-04

Oct-04

Dec-04

Feb-05

Apr-05

Jun-05

Aug-05

Oct-05

Dec-05

Mo

nth

ly D

G'd

Tra

nch

es

0

10

20

30

40

50

60

70

80

90

Mo

nth

ly D

eals

on

NW

HY CBO HY CLO IG CDO BAL SHEET

SF CDO SYNTHETIC OTHER Deals on NW

4

0

10

20

30

40

50

60

70

80

Dec-03

Feb-04

Apr-04

Jun-04

Aug-04

Oct-04

Dec-04

Feb-05

Apr-05

Jun-05

Aug-05

Oct-05

Dec-05

Mo

nth

ly U

G'd

Tra

nch

es

HY CBO HY CLO IG CDO BAL SHEET SF CDO SYNTHETIC OTHER

Source: Credit Suisse, Moody’s, S&P, Fitch Source: Credit Suisse, Moody’s, S&P, Fitch

6 For more on the REIT TruPS CDOs, please see "The CDO Strategist - Issue #8 - An Introduction to REIT

Trust Preferred CDOs", 9/30/2005, Credit Suisse CDO Research.

REIT TruPS CDOs:

leveling the playing

field for small REITs

2005 was a solid

year for ratings

performance

Early vintage SF

CDOs: the lion’s

share of

downgrades

31 March 2006


Exhibit 12: Most DGs on early vintage SF CDOs Exhibit 13: Upgrades mixed across asset classes

DG'd SF CDO Vintages in 2005

2000

23%

2001

44%

2002

29%

2004

1%2003

2%

1999

1%

Distribution of 2005 UG's

HY CLO

19%IG CBO

1%

SF/MS CDO

40%

Other

3%

HY CBO

34%

Mrkt Value

2%

Synthetic

1%

Source: Credit Suisse, Moody’s, S&P, Fitch Source: Credit Suisse, Moody’s, S&P, Fitch

Upgrades, on the other hand, were diversified among HY CLOs, HY CBOs, and recent-

vintage SF CDOs (Exhibit 13). HY CBOs & CLOs combined for about 53% of 2005

upgrades, buoyed primarily by continued amortization of CDO tranches and the

subsequent de-leveraging of the transactions, which enhanced overcollateralization (OC)

ratios. Furthermore, performance of HY corporate credits as a whole improved in 2005;

most negative performance was concentrated in specific sectors, such as autos and

airlines. Most HY CBOs/CLOs did not have significant exposure to these risky sectors.

Although, we note that many European synthetic CDOs did have exposure and were put

on watch or downgraded.

SF CDO upgrades were particularly concentrated in deals backed primarily by CRE assets

or newer vintage (post-2003) deals with high concentrations of residential mortgages

(Exhibit 14). CMBS had a solid year in 2005 as national office vacancy rates improved,

upgrades outpaced downgrades by eight to one, and many seasoned bonds were

defeased, justifying upgrades of CRE CDOs holding these assets.7 On the mezzanine

ABS CDO side, recent vintages warranted upgrades as prepayment speeds for 2003 and

2004 vintage residential mortgage pools were among the highest over the last ten years

while delinquencies were among the lowest.

Exhibit 14: Distribution of 2005 Upgrades by vintage & asset class

Vintage HY CBO HY CLO IG CBO SF/MS MV Synthetic Other Total

1996 3 2 5

1997 14 2 16

1998 21 14 6 41

1999 40 14 5 3 62

2000 24 7 1 4 36

2001 1 1 2 11 3 1 19

2002 5 33 1 39

2003 13 42 55

2004 34 34

Total 103 58 3 125 6 3 9 307

Source: Credit Suisse, Moody's, S&P, Fitch

7 For more on CMBS performance, please see "CMBS Market Watch Weekly", 12/16/2005, Credit Suisse

CMBS Research.

Upgrades were

diversified

CRE CDOs and

recent SF CDOs

backed by

RMBS/HEL were

upgraded

31 March 2006


Finally, 2005 saw a record number of ratings withdrawals – 374 tranches across 120 deals.

HY CBOs/CLOs accounted for over 75% of all withdrawn ratings by tranche count. Nearly

all withdrawals were because of tranche redemptions or whole deals being called pursuant

to the optional redemption from equity holders. Asset appreciation and the lack of

alternative investments, both due to tightening spreads over the last two years, nurtured

an environment for deal redemptions.8

CDO spreads: defining stability For most of 2005, CDO spreads remained stable after tightening significantly in 2004

(Exhibit 15, Exhibit 16). As we’ve discussed, supply was exceptionally strong last year;

demand from yield-seeking investors was equally robust. Volatility from hedge fund related

buying of protection on ABS was the straw that broke the camel’s back, forcing

subordinate spreads to widen for mezzanine ABS CDOs during Q4. In particular, BBB

mezzanine ABS CDO spreads widened 80 bps within a few weeks, with one pricing seen

at L+400 bps late last year.

Exhibit 15: 2005 select AAA CDO Spreads Exhibit 16: 2005 select BBB CDO Spreads

26

29

33

20

25

30

35

40

45

Dec-04 Feb-05 Apr-05 Jun-05 Aug-05 Oct-05 Dec-05

HY CLO AAA ABS M Z AAA

CRE AAA BTRUP AAA

180

350

290

200

220

150

200

250

300

350

400

Dec-04 Feb-05 Apr-05 Jun-05 Aug-05 Oct-05 Dec-05

HY CLO BBB ABS M Z BBB

ABS HG BBB CRE BBB

BTRUP BBB

Source: Credit Suisse Source: Credit Suisse

While most new issue Triple-A spreads have converged, spread tiering exists down in

credit. As shown above, AAA spreads for SF CDOs and HY CLOs converged at L+25 –

27 bps for most of the year before widening slightly towards year-end. BBB SF CDOs

ended 2005 about 170 bps wider than BBB HY CLOs. The new product premium

associated with TruPS CDOs is mostly gone (with the exception of REIT TruPS CDOs)

and there has also been a divergence between high-grade and mezzanine ABS CDO

spreads at the subordinate level.

8 For more on CDO optional redemptions, please see "The CDO Strategist - Issue #2 - When's The Best

Time to Call? - Optimal Timing of CDO Calls and Relative Values", 5/31/2005, Credit Suisse CDO Research.

CDO spreads mostly

unchanged, until Q4

Spread tiering exists

among products

down the credit

spectrum

31 March 2006


Our top 5 list for 2006 In the following section, we provide our top 5 issues CDO investors should be mindful of in

2006. Each issue may very well be an individual research topic, and so we’ve restricted

our comments to brief highlights.

#1. Will the credit cycle reach a turning point in ‘06?

The consensus of crystal balls says “no”.9 While some may believe the outlook is murky,

nearing the tipping point, we believe evidence supports the contrary. Aggregate corporate

credit quality is set to stay robust as profits remain high, cash flows remain robust, and

external funding requirements remain minimal, barring any unexpected global financial

distress. As with 2005, corporations continue to be flush with cash as investments remain

very subdued relative to the current level of corporate profitability and cash flow.

Additionally, the economic picture continues to look robust following a solid 2005. US real

GDP is expected to grow at 3½%, not far from 2005’s projected path of 3¾%, although the

composition of growth is expected to diverge from 2005’s.10 Growth will shift from the

housing market and the consumer to the corporate landscape and the global economy.

While there are risks and uncertainties on the horizon, including fears of amplified LBO

activity, Fed tightening concerns, and increasing leverage, we believe the favorable

corporate credit and economic environments outweigh these fears and should keep

defaults low with a modest rise, if any.

#2. Where is the housing market headed?

Down, but not out. While we do expect moderate cooling in the housing market and some

localized distress, we do not expect anything close to a national drop-off in activity. Signs

of a slowdown have emerged as unsold home inventories have increased, and existing

home sales have leveled off. Home price appreciation slowed in Q3 2005 as YOY HPA

declined to 12.02% from 14.01% in Q2 2005. 11 Additionally, greater leverage in the

mortgage market for new home buyers, coupled with rising interest rates and increasing

regulatory scrutiny of affordable mortgage products, raises the risk of a regional downturn

absent an economic trigger.12

However, with all this said, several key components supporting consumer credit should

keep the consumer in the game. As we mentioned in the previous point (#1 above), the

US economy is expected to remain healthy, with GDP growing at 3½% in 2006, which

should continue to support job growth. With the job market remaining robust, and 30-year

fixed mortgage rates essentially unchanged from when the Fed started tightening in 2004,

growth in disposable income and wealth gains will support the consumer’s ability to

service his/her debt in 2006.

#3. Will 2006 be another record year for the CDO issuance?

Given that 2005 turned out to be a blockbuster year for CDO issuance and the majority of

market participants expect the supply of many cash collateral markets for CDOs – such as

home equities – to fall in 2006, it seems reasonable to expect a drop in CDO volume.

9 This section makes significant references to "2006 US Credit Outlook", 12/14/2005, Credit Suisse US

High Grade Credit Team, and "Leveraged Finance Strategy Outlook 2006", 1/16/2006, Credit Suisse Global Leveraged Finance Strategy and Portfolio Products Team. 10

"US Economics Digest: Forecast Review - 2006 Outlook", 12/16/2005, Credit Suisse US Economics Team 11

HPA stands for Housing Price Appreciation, which is calculated based on the housing price index. 12

"Quarterly Home Price Update: Is It Different This Time?", Credit Suisse ABS Research, 12/30/2005

Corporations remain

flush with cash, as

investments remain

low and profitability

remains high

Defaults should stay

low, despite some

risks &

uncertainties

We do not think a

national housing

bubble exists

31 March 2006


However, we see the synthetic market as the “wild card” – both in terms of synthetic CDOs

and synthetic buckets in cash deals (including hybrid deals). The growing usage of

synthetics will mitigate the difficulty of collateral sourcing for cash deals and the

advantages of synthetic CDOs will push their volumes much higher in 2006.

On a product-specific basis, developments in CDS of loans and ABS continue to

proliferate the market. A standardized loan CDS template is expected in Q1 2006 to

address the prepayment exposure of loans. Like the application of CDS on ABS, a

standardized, tradable loan CDS could expand the synthetic loan market exponentially – a

trend certainly worth watching. On the ABS side, the CDS market will also be extended to

other ABS asset classes such as autos, credit cards, and CDOs. Furthermore, the launch

of the ABX index (on January 19th, 2006) could provide another dimension to taking on or

hedging risk in ABS.

Additionally, a proposed amendment to FAS 140, which changes the accounting treatment

of synthetic CDOs, could impact the market dramatically in 2006. Many investors are

unable to participate in synthetic CDOs because of accounting bifurcation and mark-to-

market (MTM) treatment of the CLN issued by the synthetic CDO. The amendment would

align the credit risk exposure of the CLN issued by the synthetic CDO, thereby mitigating

the bifurcation and MTM issues and reducing the P&L volatility associated. The proposal

could greatly expand the investor base for synthetic CDOs.

On the demand side, given a continuing low yielding and low volatility environment, CDOs

still offer very attractive spreads and thus we expect the appetite for CDO products to

remain strong. In Exhibit 17, we list our issuance projections for each CDO sector and

briefly comment on their justification.

Synthetic CDOs

could make the

difference in 2006

FAS 140 proposal

could significantly

expand synthetic

CDO investor base

31 March 2006


13 Please see "Leveraged Finance Strategy Outlook 2006", 1/16/2006, Credit Suisse Global Leveraged

Finance Strategy and Portfolio Products Team

Exhibit 17: 2006 Issuance Forecast

CDO Sector 2005 Actual Change (%) 2006 Projection Reason

HY CLO $46.4 flat $46Expect institutional loan issuance to decline.13 Collateral spreads to remain tight or

range-bound, chipping away at CLO equity returns.

MML CLO $11.3 up 10% $12.5 More hedge fund participation in lending to small & medium sized companies.

Mezzanine SF CDO $24.6 down 15% $20.9Decline in supply as refinance activity shrinks; profit margins for lenders tight;

slowdown in home price appreciation.

High Grade SF CDO $51.0 down 10% $45.9 Same as above, except decline is less as there are more senior HEL bonds.

CRE CDO $15.7 up 15% $18.1Real estate fundamentals remain in place with property values in good shape; Low

long-term interest rates; Strong demand for CRE assets.

CDO^2 $5.5 up 10% $6.0Record CDO issuance since 2004 drives supply; improving technology to analyze

complexities of CDO^2 transactions; rise of CDO hedge funds.

TruPS CDO $9.7 up 10% $10.6Early vintage bank TruPS reaching 5-yr non-call; proliferation of REIT TruPS as

stand-alone CDOs or as collateral for hybrid TruPS CDOs.

Market Value CDO $5.1 up 10% $5.6 Evolution of market value technology and better management flexibility.

Synthetic CDO $16.6* up 50-100% NA*

New product innovation: ISDA template for CDS on loans expected, ABX Index,

continued growth of CDS on ABS and expansion into other ABS areas such as

consumer products, CDOs, etc.; amendments to FAS 140 should expand demand

significantly for synthetic CDO.

EM CDO $1.1 up 10% $1.2EM economic growth to continue in 2006, especially in China; most EMs have tame

inflation.

Other $3.0 flat $3.0

Total $188.0 up 5-12% $210.0While SF CDO issuance, the lion's share of the US CDO market, is expected to

drop, we believe the CRE CDO and synthetic markets will make up the difference.

*Because of the private nature of many synthetic transactions, issuance figures reported may be much smaller than actual.


31 March 2006


#4. Manager Selection: Raising the bar

2005 saw 170 managers (of which 58 were new) crowd the CDO market; that’s almost one

manager every two days! A breakdown by the top 3 asset classes reveals 75 managers in

the CLO space, over 65 managers in the ABS CDO market, and 23 managers in the CRE

CDO arena.14 Managers entered the CDO space from three directions: new managers

entered the accommodative environment in droves, while seasoned managers returned to

their assets of expertise and/or ventured into new asset classes.

With little to no spread tiering among managers in the past year, it seems the market has

not priced in the benefit of a strong manager or the risk of a poor one. Robust demand for

CDOs coupled with a benign credit and favorable housing environment, helped drive this

trend. We believe that so long as these factors remain in place, tiering will not occur. As

we discussed earlier, we do not believe the credit cycle will be turning in the near term nor

will there be a dramatic collapse in the housing market. However, with managers dipping

further down in credit for spread, it will become more crucial to be selective in picking

asset managers. But how do we accomplish this? What should we look for?

The simple response we often hear from investors is to pick managers who have

“managed through the cycle.” This limits investors to a handful of managers in the market

for longer than five years. While the HY CLO market has been active over a relatively long

timeline, we count only 24 of the 75 managers who issued a deal in 2005 as having issued

a deal between 1998-2001, too. The ABS CDO market has only been around since 2000,

and pre-2003 deals are significantly different from deals post-2003.

For any manager, we think investors should ask the following questions:

1. How important is the CDO business to this manager? I.e., is this their primary

business? And if not, how does it compared with its other businesses? We prefer

repeat and bigger managers, or small/new managers with extensive experience

at their previous employer – a bigger manager with a good track record.

2. What is the manager’s background and core experience?

3. What is the manager’s investment philosophy and does it fit your criteria?

4. How does the manager deal with distressed/defaulted assets? What is the

manager’s experience with work-outs in the case of certain assets?

5. Does the manager re-rate each asset or follow rating agencies' ratings? How

good is the manager's internal credit monitoring system?

6. For SF CDOs, does the manager have an internal rating system for originators

and servicers?

7. Does the manager have a strong credit research team?

8. What kind of systems/modeling software does the manager use?

In addition, prudent investors should also request historical equity returns, ratings

performance, and lists of past credit risk sales and portfolio purchases where available.

Furthermore, we think structural features like key-man provisions help protect investors.

14 Overlaps exist, i.e. there may be a manager that manages both CLOs and ABS CDOs. This is counted

once for each asset class managed.

170 total CDO

managers in 2005

Little to no spread

tiering among

managers

Having “managed

through the cycle”

is not the only

criteria

31 March 2006


#5. Rating Methodologies

Recently, each rating agency released a revision or improvement to its rating methodology.

Just to name a few: Moody’s introduced its Correlated Binomial Model for SF CDOs in

September 2005; S&P released a new version of CDO Evaluator (see Strategy section);

and Fitch made accessible the ability to model Leveraged Super Senior structures in its

VECTOR model. We expect this trend to continue in 2006 as more new products and/or

structures emerge and more empirical data and better modeling techniques become

available. It is critical for all participants to stay on top of any change to the “rules”, as

minor changes in criteria or assumptions could have major consequences to primary and

secondary markets.

Closing thoughts In 2005, the US CDO market reached a new milestone by surpassing all issuance

expectations, breaking new ground in synthetics and in its applicability as an arbitrage,

balance sheet, and term financing tool. Looking towards 2006, the credit and economic

environments appear set for another solid year, while the housing market, with glimpses of

softness, may still have some steam left. The stage is set for the synthetic CDO market to

make waves in the US and we reiterate the importance of diligent manager selection as

spreads remain range-bound and managers dig deeper for yield.

As we make our way to the “back nine” of 2006’s CDO market, we look forward to being

your caddie along the way, helping you navigate on the fairway.

31 March 2006


Chapter 1. Structured Finance CDOs

31 March 2006


Structured Finance CDO Primer15

Overview First introduced in 1999, in just six years, structured finance CDOs (SF CDOs) have

become a mainstay of the US CDO market, comprising of about 51% of the market, or

$91bn in 2005 (Exhibit 18 and Exhibit 19).

Structured Finance Securities (SFS) have historically exhibited more stable credit

performance. About 90.47% of triple-B SFS, including ABS, CMBS, CDO and RMBS,

remained in the same rating category over the course of a year, versus 88.25% for

triple-B corporates, and 82.26% and 81.66% for double-B and single-B corporates,

respectively.16

SF CDOs opportunistically capitalize the liquidity premium on subordinate SFS.

Historically, subordinated classes, e.g., triple-B SFS, carry more liquidity premium than

corporates, in part due to relatively small tranche sizes, a limited investor universe, and,

sometimes, structural complexity. Thus SFS have offered wider spread than corporates.

SF CDOs, largely a buy and hold vehicle with limited collateral trading, are ideal for

capitalizing the collateral’s liquidity premium. By securitizing and tranching a pool of mezz

and subordinated SFS, SF CDOs create higher-rated and likely more liquid senior classes,

and in the meantime, leverage collateral liquidity premiums to generate attractive equity

returns. At times, SFS spreads have widened out due to market technicals as opposed to

heightened credit risk, such as the reduced liquidity in the aftermath of 9/11 in 2001.

Ramping-up SF CDOs during times like these may result in greater collateral spread and

greater excess spread, enhancing equity returns.

Historically SF CDO bonds have traded cheap to other CDO products. SF CDOs

have generally priced wider than most other structured products, resulting in better relative

value for investors in SF CDOs. This is due, in part, to the relative newness and

complexity of this product. For example, BBB SF CDO tranches still offer significant

spread pick-up versus other CDO types and structured products in general.

Other factors propelling the growth of SF CDOs include increased participation from a

variety of issuers. Over time, SF CDOs have attracted a new genre of issuers, including

hedge funds and specialty structured finance investment companies. In seeking viable-term

funding in the aftermath of the 1998 LTCM/Russia crisis, hedge funds such as TCW,

Ellington Capital, Vanderbilt Capital and Maxim Advisory have repetitively utilized SF CDOs

to secure more stable term funding. To achieve more efficient funding, specialty structured

finance companies such as SFA, C-BASS, GMAC and Fortress Investment also have

become repeat CDO issuers, mainly in real estate CDOs. In addition, money managers

such as PIMCO, Rabo, Independence, Oppenheimer Funds, and Deerfield are repeat CDO

issuers, and have continued to broaden their product types, including issuing SF CDOs.

Basic Terms of SF CDOs To understand how SF CDOs work, we first illustrate a typical CDO timetable (Exhibit 18).

Similar to cash flow HY CDOs, non-static pool cash flow SF CDOs often have a ramp-up

and reinvestment period, which often falls within a non-call period. For example, the

Pacific Shores CDO, which closed on June 27, 2002, had a 60-day ramp-up period, a 3-

year reinvestment period and 4-year non-call period.

15 This section was originally written by Neil McPherson, Helen Remeza, and David Kung, March 2003;

sections in bold have been updated as of March 2006. 16

Sources: “Structured Finance Ratings Transitions: 1983-2005” February 2006, Moody’s Investors Service.

Four main factors

contributed to the

growth of SF CDOs

1) Historically stable

SFS credit

performance

2) SF CDOs

capitalize liquidity

premium on sub

SFS

3) SF CDO bonds

are higher yielding

4) An increased

issuer participation

A typical timetable

31 March 2006


Exhibit 18: A typical SF CDO timetable

Ramp up Reinvestment Pay-down period Start of portfolio warehousing Non-call period Call period

3 to 5 years

Pricing/Closing Legal MaturityStep-up or Auction Call

date

8 to 35 years

3 to 5 years


Exhibit 19: Deal Terms – Pacific Shore CDO terms Basic Terms

Deal Type: Cash Flow ABS CDO

Assets: ABS, CMBS, RMBS, CDO (See Exhibit 6 for detail)

Closing Date: June 27, 2002

Target Par Amount: $705.5 million (Upsized from $600mm)

Ramp-Up Period: 60 days from closing (90% ramped up at closing)

End of Reinvestment Period: May 2005

Interest Payment Frequency: Quarterly, beginning October 3, 2002

Reinvestment Period: 3 years

Non-Call Period*: 4 years

Auction Call Date: 10 years

Maturity Date May 2037 (35 years)

*After the call date, all of the Class A Notes, Class B Notes and Class C notes may be called by a majority vote of the holders of

the Preference Shares.

Pricing Table

Tranche Size % of Deal

Rating

(Moody's/S&P/Fitch)

WAL

(Years) Coupon

A $532,000,000 75% Aaa/AAA/AAA 5.6 L + 47

B-1 $96,000,000 14% Aa2/AA/AA 10.0 L + 89.5(1)

B-2 $16,000,000 2% Aa2/AA/AA 10.0 L+75

C $28,000,000 4% Baa2/BBB/BBB 7.5 L + 230

PS-CL1 $21,500,000 3% Ba3/BB-/BB-(2) -- --

PS-CL2 $7,000,000 1% Ba3/BB-/BB-(2) -- --

Pass Through Notes (3)

$5,000,000 1% -- -- --

(1) This is the max rate for this auction rate note class.

(2) This is a principal only rating.

(3) The class is a combo note with a small equity participation.

Source: Credit Suisse, IFRMarkets, MCM & Bloomberg

The deal pays 25bps of senior and 25bps of junior management fees to PIMCO, plus an

incentive fee of 20% of equity cash flows after a 12.5% IRR is achieved. 17 The fee

structure is negotiated up front, and often differs from one deal to another. While a senior

fee supports a manager’s ongoing operations, a junior and/or an incentive fee helps to

align the manager’s interests with those of subordinate holders.

The pricing table above indicates a 4% enhancement to triple-B. Rating agencies generally

permit higher leverage for SF CDOs than for HY CDOs, mainly because the average

collateral credit quality for SF CDOs is investment grade, i.e., far less likely to default than

high yield bonds.

17 “Senior” fees are paid at the top of the waterfall and are thus more likely received by the manager than

“junior” fees, which are paid further down at the bottom.

31 March 2006


Compared to HY CDOs, the deal term of SF CDOs differs in several ways. The legal final

of SF CDOs is often longer due to long collateral, i.e., a 30~35-year legal final versus 12

years for HY CDOs, though the expected final of SF collateral is actually far shorter than

its legal final. In addition, for arbitrage-driven trades, the ramp-up period of a SF CDO may

be longer (typically ranging from 3~6 months), partly due to a thinner subordinate SFS

market. Some key structural innovations addressing these issues include step-up coupons,

auction calls, put options, turbo mezz and dynamic funding.

Collateral Analysis Once a SF CDO is closed, sales proceeds are used to fund the collateral portfolio. At closing,

non-static investment-grade SF CDOs are often partially funded (typically at least 50%) with

a “warehouse line” and with a required amount invested in investment-grade SFS.

To help in tracking a dynamic pool of credits (most CDOs are managed pools, which allow

for limited reinvestment and trading), rating agencies have established a set of criteria to

maintain the consistency of collateral as the deal evolves, versus its initial pool. Some of

the key guidelines include trading limitations, diversity and min/max collateral spread

requirement, rating, coupon, average life, name or industry or issuer or servicer

concentration.

In most SF deals (as SF CDO collateral), the issuer as servicer is responsible for loan

payment collections and delinquent management. In consumer finance transactions,

however, servicer troubles may lead to declining servicing quality, and this can potentially

impact bond performance. To mitigate this risk, SF CDOs often limit exposure to single

servicer, in a sense similar to HY CDOs, which limit exposure to single credit. A lower-rated

servicer is subject to a tighter limit. For example, in PIMCO’s Pacific Shores CDO, the

maximum concentration for below ‘A-‘ or ‘A3‘ or ‘S2’ rated servicers is 7.5%, while the limit is

10.5% for ‘A3’ or higher but below ‘Aa3’ rated servicers, and 15% for ‘Aa3’ or higher rated.

To limit sector concentrations and quantify portfolio diversification, rating agencies have

developed their own measures of correlation. A detailed discussion on this topic is beyond

the scope of this report. We discuss briefly Moody’s methodology. Moody’s original

diversity/sector score reflects the impact of sector concentration and correlated defaults.18

For example, a well-diversified portfolio (i.e., the asset default correlation is lower) is less

susceptible to name-specific risk and often achieves a higher score. Compared to HY

CDOs, typically SF CDOs are assigned a lower diversity score, i.e., a 20 Moody’s diversity

score vs. over 40 for HY CDOs. Moody’s have revised their methodology in September

2005 and introduced their new “correlated Binomial Expansion Technique (BET)”.19 The

new model uses asset correlations, rather than default correlations in diversity score

calculation, in Moody’s CDOROM model, which is a model to derive the loss distribution of

the underlying collateral pool.

SFS Rating Stability One particularly appealing aspect of SF CDOs is that historically SF collateral has

exhibited better rating stability than corporate bonds. For example, About 90.47% of

triple-B SFS, including ABS, CMBS, CDO and RMBS, remained in the same rating

category over the course of a year.

18 Moody’s Approach to Rating Multisector CDOs, Moody’s, September 15, 2000.

19 Moody's Modeling Approach to Rating Structured Finance Cash Flow CDO Transactions, Moody's,

September 26, 2005.

The deal term of SF

CDOs differs from

HY CDOs

Key portfolio

guidelines

SF CDOs also

impose servicer

concentration limits

Diversity/sector

scores were

originally used to

measure diversity

diversification

Historically SFS

exhibited better

rating stability…

31 March 2006


Corporate guaranteed, wrapped bonds and SF deals backed by “lumpy” collateral may be exposed to more concentrated risk (event risk). Event risk is more significant

in less diversified SFS deals, including airline-linked aircraft lease deals (EETC and to

some extent pooled ETC) and CMBS large loan deals, and in corporate guaranteed

structured securities. For example, some non-IG (investment grade) SFS are guaranteed

by originators, who are typically IG at origination. Should collateral credit quality erode

and impact the credit worthiness of the bond, these institutions have a contractual

obligation to guarantee the bond payments. The credit risk of these bonds is thus linked to

a single company, and this introduces event risk. For example, there were 32 defaults in

the manufactured housing (MH) market in 2002; all of them were corporate guaranteed by

either Oakwood or Conseco, both of which filed for bankruptcy.

SFS Recovery Value Another appealing aspect of SFS historically has been that they have generally achieved

higher recovery values than corporate bonds, in part thanks to the secured nature of select

products such as RMBS and CMBS. These are mainly backed by hard assets, the “bricks

and mortar” type. Also while corporate bonds, once defaulted, usually stop receiving

payments, SFS typically continue to receive cash flow for quite some time after default.

Based on all defaults, S&P’s recovery study of September 2002 suggests RMBS and

CMBS achieved a high recovery rate, 60% and 83%, respectively, while the ABS average

recovery rate reached 45% (Exhibit 20). 20 Across the new defaults observed during

June01~June02, S&P suggests RMBS and CMBS recovered about 98% and 87%,

respectively, while the average ABS recovery rate reached 62%. Also, excluding the

securitizations of charged off credit card receivables (including the three fraudulent CFS

deals) and synthetic deals, ABS recovered at a higher rate (77%).

We should note that the number of 'D' rated classes increased by 64 to 178 over the one-

year period ending June 2002 from the previous 114 over a 15-year period, in part due to

the currently weak credit environment. Nevertheless, it appears that SFS continue to

exhibit low defaults, as only 178 SFS classes were ‘D’ rated across 18,500 US SFS S&P-

rated classes over the past 16 years.

We offer the caveat that S&P’s SFS default definition and recovery calculation requires

careful interpretation. For example, S&P’s “maximum possible recovery” assumption implies

some optimism, as its recovery value calculation is based on the current cumulative loss

without giving consideration to likely future losses (from the calculation date onward to the

final maturity). However, S&P considers a bond experiencing an “interest shortfall” (S&P

defines this as missing a dollar of the scheduled interest payment) to be a defaulted security.

The implication of this is that should the erosion continue, it may lead to a lower future

recovery; or should the interest shortfall be cured, it may imply a 100% future recovery. That

having been said, we think the overall SFS recovery rate is respectable thus far, and this has

positive implications for SF CDO investors.

20 S&P structured finance recovery study, September 2002.

Guaranteed SFS are

more exposed to

more concentrated

risk

High recovery for

real estate SFS is

partly due to hard

collateral or “B&M”

Exhibit 20: S&P Suggests SFS Exhibited Relatively High Recovery Rates

RMBS CMBS ABS

Original New Defaults All Defaults New Defaults All Defaults New Defaults All Defaults

Rating Count Recovery Count Recovery Count Recovery Count Recovery Count Recovery Count Recovery

AAA 3 96% 2 93% 2 93%

AA 3 91% 20 75% 1 89%

A 2 93% 3 66% 1 0% 7 71% 17 38%

BBB 4 83% 17 67% 10 37% 10 37%

BB 4 94% 15 67% 8 97% 9 97% 5 65% 7 46%

B 5 75% 30 36% 10 98% 16 79% 3 100% 5 60%

CCC 1 100% 1 100%

All / Avg. 19 87% 89 60% 18 98% 27 83% 27 62% 41 45%

*The inception of the study is 1985, 1985 and 1978 for ABS, CMBS and RMBS, respectively.

Source: S&P, Credit Suisse

31 March 2006


Structural Considerations Similar to other CDOs, SF CDOs are often structured with coverage tests such as over-

collateralization (OC) and interest coverage (IC). Should actual OC/IC fall below the test

level, usually excess spread is first used for reinvestment or paydown of senior notes to

bring the test back into compliance. An initial OC cushion is defined as the difference

between the initial OC and OC test level, and it varies by deal, tranche rating and structure.

While OC tests are often the first line of defense for senior note holders should undue par

erosion occur, IC tests generally are structured with a much larger cushion.21

Besides the coverage tests, there are a few unique features in long legal maturity CDOs

like SF CDOs. To shorten the average life of liabilities, SF CDOs often build in some

unique structural features, such as a liability coupon step-up, auction calls, and turbo pay-

downs. Sometimes to shorten average life, on a secondary basis, CDO investors may buy

a put option with a customized exercise date. We will explain some of these features in

more detail in the discussion that follows.

Step-up Coupon. For example, in both the DASH and Bleecker CDOs, in year 12, the

mezz coupon will step-up by a large margin, i.e., 500bps, giving more incentive for equity

holders to call the deal as this drains available excess spread cash flow. Should the

collateral credit profile remain healthy, this feature will likely result in a shorter average life

for liabilities. In other words, managers are likely to retire a portfolio of seasoned and well-

performing SFS, as managers will try to avoid a steep penalty to equity returns due to the

coupon step-up. The call may not be attractive to the manager, however, if the pool

becomes credit impaired or trades well below par.

Turbo Mezz. During the reinvestment period (typically the first three to five years) of a SF CDO,

some deals utilize a portion of excess interest, i.e., after a predetermined equity dividend

hurdle or “capped equity return”, to amortize the mezz tranche before the inception of a

paydown period. This is known as the “turbo” mezz or “turbo” triple-B feature. This allows the

mezz OC cushion to slowly build up and enhances mezz. From the senior’s perspective, this

replaces subordination with OC. In addition, it enhances available excess spread as more

expensive mezz liabilities are paid down. Furthermore, the early principal payback shortens the

average life of the mezz tranche ultimately and may also reduce the expected principal losses

in high default scenarios. Across the 10 SF CDO universe we track, we found an average 8%

annualized turbo paydown during an average seasoning of 10 months.22

Auction Call Redemption. This is another common feature with long maturity CDOs.

Barring any unexpected credit deterioration in the pool, it is likely that an auction call will

be in-the-money. Typically, after the tenth anniversary of a transaction, the CDO trustee is

required to conduct auction calls on a regular basis, soliciting bids on the collateral pool

from interested parties to retire the notes. The auction dates often coincide with payment

dates. If the notes have not been redeemed in full prior to the tenth year, there will be a

continual mandatory auction call process until all notes can be redeemed in whole at

once.23 In addition, after year-10, typically all excess spread will be used to pay down

liabilities. Auction call redemptions are likely to enable an early return of principal, and, as

such, shorten the average life of CDO liabilities. At the auction call date, two things are

21 This is partly due to the forward interest rate curve assumption and to some extent other tests such as

the weighted average spread (WAS) test and weighted average coupon (WAC) test have already addressed the interest coverage. 22

We note that the turbo mezz is a recent feature, thus we only have limited history. Also see our special report “Turbo triple-Bs in ABS CDOs”, published in Nov. 2002. 23

The trustee, in behalf of the collateral manager, certifies a successful auction if the sale proceeds from the collateral which, together with the balance of all eligible investments and cash in the accounts (other than the Hedge Counterparty Collateral Account and the Cash Flow Swap Counterparty Collateral Account), will be the total redemption amount of notes (often including accrued/deferred interest), plus the greater of 1) zero, and 2) the aggregate initial purchase price of the equity, minus the total cash distributions on the equity.

Coverage tests

Several structural

features aimed to

shorten bond WAL

Step-up coupon

motivates manager

to call the deal

Early paydown of

mezz to build OC

Auction call

31 March 2006


likely to have happened: 1) the collateral would have seasoned, i.e., stabilized with respect

to credit (and prepayments if applicable), and shortened its remaining average life,

possibly being sold at tighter spreads (or higher prices); and 2) the triple-B would have

been partly paid down from debt turboing, reducing the amount of outstanding CDO

liabilities. Both of these may result in an in-the-money auction call; i.e., the value of the

collateral pool being greater than the value of the liabilities. Separately, after the regular

non-call period expires, equity holders are increasingly likely to call the deal, as the CDO

may have de-levered from mezz turboing and collateral paydown, which reduces the

leverage and arbitrage. We caution that should collateral credit deteriorate, both the

auction call and the regular call become less likely to be in-the-money.

Finally, SF CDOs sometimes require longer ramp-up, partly due to limited collateral supply.

Some structural remedies include a dynamic funding mechanism such as variable funding

notes (VFNs) or delay draws, and the issuance of additional notes.24 Dynamic funding

alleviates negative carry during the ramp-up period, reducing time pressure in acquiring

the target portfolio. In some dynamic funding structures, the liability outstanding amount

increases in sync with the CDO portfolio balance through the issuance of additional notes,

subject to pre-negotiated rating agency conditions. In other cases, CDO liabilities are

structured as revolving credit facilities such as VFNs or delay draws, which offer more

flexibility in funding and can be drawn down at short notice. The VFNs can be particularly

attractive to short-term LIBOR funders like ABCP conduits.

Besides the above unique features, SF CDOs continue to be fine-tuned. Some recent

structural refinements include applying a par haircut for deep discount purchases, and

triple-B or below rated collateral, and the rapid amortization of mezz SF CDO tranches.

Arbitrage and Performance Indicator For arbitrage-driven cash flow deals, a key consideration is the “arb” level. To track the

“arb” or “excess spread” for SF CDOs or Multi-sector CDOs, we introduced a tool called

Multi-sector Arbitrage Pointer (MAP) (Exhibit 21, Appendix I). It is similar to the commonly

cited “HY-CBO arb,” except MAP reflects the market conditions specific to the structured

finance collateral contained in SF CDOs. It captures both the excess spread between

assets and liabilities, and the implications for the return on equity (ROE) in one number. In

other words, MAP’s utility lies in providing some direction on new issuance volumes and

expected ROE. If higher excess spread is indicative of less liquidity rather than poor credit

quality, a higher MAP in combination with higher leverage, enhances ROE. For example,

a 150bp MAP in combination with a 20 leverage ratio indicates a 30% zero default ROE.

Under realistic default scenarios, strong return on equity generally spurs CDO issuance.

Exhibit 21: Multi-sector Arbitrage Pointer (MAP)

66

0

30

60

90

120

150

Oct-

01

Jan-

02

Apr-

02

Jul-

02

Oct-

02

Jan-

03

Apr-

03

Jul-

03

Oct-

03

Jan-

04

Apr-

04

Jul-

04

Oct-

04

Jan-

05

Apr-

05

Jul-

05

Oct-

05

Jan-

06

Exce

ss s

pre

ad

in

bp

s


24 “Dynamic funding in cash flow arbitrage CDOs,” Fitch, February 2003.

Dynamic funding

allows for a longer

ramp-up

Credit Suisse’s MAP

monitors SF CDO

arbitrage

31 March 2006


Key catalysts for sustainable stable SF CDO performance include healthy ABS credit

behaviour, continuous CDO structural refinements and prudent execution. As with all CDO

products, we advocate a close scrutiny of collateral, structure and manager. Compared to

most other structured finance securities, SF CDO bonds remain higher yielding across the

credit spectrum. We like well-structured SF CDOs backed by carefully selected collateral

and managed by reputable issuers seeking a long-term market presence.

Appendix I. Calculation of MAP and LAP

Our Multi-Sector Arbitrage Pointer (MAP) and Leveraged-Loan Arbitrage Pointer (LAP)

reflect the excess spread between assets and liabilities in structured finance/multi-sector

(SF/MS) CDOs and Collateralized Loan Obligations (CLO), respectively. An increasing

MAP or LAP suggests higher arbitrage and thus higher return on equity, making it more

attractive for issuers for potential new deals.

We derive MAP and LAP using two static “generic” collateral portfolios with defined sector

and rating allocations typically found in CDOs. These portfolios are adjusted over time to

reflect the actual asset allocation better across different vintages. The chart below shows

the current asset allocation for MAP and LAP calculation:

Exhibit 22: MAP/LAP Collateral Composition*

MAP

HEQ

60%

CC

2%

CMBS

5%

Alt-A

10%

CDO

10%REIT

2%Corp.

1%Jumbo

RMBS

10%

LAP

B Corp.

Bonds

4%BB Inst.

Loans

32%

B Inst.

Loans

63%

BB Corp.

Bonds

1%


We note that the collateral loan spreads are based on CS’s Leveraged Loan Index, high

yield bond spreads are based on CS’s High Yield Index, and investment grade corporate

bond spreads are largely based on CS’s Liquid U.S. Universe Corporate

Index (LUCI), where only larger and/or more

liquid names are included.

The cost of funding is approximated by the

weighted-average liability spreads. Please

refer to the table on the left for the capital

structures used.

We then take the difference between the

aggregate asset spread and the cost of

funding, less 50 basis points for management

fees and other costs.

Closing remarks

Exhibit 23: MAP/CAP Liability Capital Structure

MAP LAP

AAA 80%* 74%

AA 10% 5%

A 1% 7%

BBB 5% 6%

Equity 4% 9%

* 70% senior AAA, 10% junior AAA

Source: Credit Suisse,

31 March 2006


High Grade SF CDO Primer: Q&A25

In this section, we introduce high grade cash flow structured finance (SF) CDOs. We

answer some key questions related to collateral, structure, and investor considerations. In

general, high grade SF CDOs are backed by high quality diversified structured finance

pools with an average rating of double-A. The short-term notes issued by these deals offer

an attractive spread pickup vs. other money market alternatives, while they are protected

by structural subordination, and to some extent credit enhanced by the put provider. Also,

the liquidity of the short-term notes is enhanced by the put provider and the relatively large

size of these notes.

What are high grade SF CDOs? The collateral for high grade SF CDOs is mainly highly rated structured finance paper,

which is typically rated at least single-A with an average rating of double-A. Some

common structured finance sectors included in the collateral pool are real estate ABS

(Resi B&C/HEL), CDO, CMBS and other ABS. High grade SF CDOs can be done on a

funded (cash flow) or synthetic basis. The decision between a cash flow vs. a synthetic

execution is mainly driven by considerations such as funding, balance sheet and/or

consolidation considerations. In this piece, we focus on cash flow high grade SF CDOs.

What does a typical collateral pool look like? The average collateral credit rating for high grade SF CDOs is typically double-A, while no

assets are rated below single-A (Exhibit 24). The deals are primarily backed by floating-

rate collateral, and therefore there is very little asset/liability mismatch assuming most

liabilities issued are floating rate. There are also portfolio concentration limits for private

securities, PIK bonds, synthetic securities as related to counterparty exposure, and for

previously troubled sectors such as MH and some esoteric ABS. If the assets are highly

rated, i.e., triple-A, there can be no limit on exposure. The concentration limit also varies

across collateral rating and servicer rating. For higher rated collateral, the concentration

limit is higher, i.e., 2%~2.5% for triple-As, while for lower rated collateral, the limit is lower,

i.e., 0.5% for single-As (Exhibit 25). Similarly, while generally servicer limits are between

7.5% to 10%, a deal may have as much as 12.5% exposure to a higher rated servicer and

7.5% to a lower rated one (Exhibit 25).

Collateral pools can be static or revolving, with revolving pools often allowing for 10% to

15% per annum of discretionary trading.

25 This section was originally written by Neil McPherson, Helen Remeza, and David Yan, July 15, 2004.

31 March 2006


Exhibit 24. Collateral characteristics for four recent of high grade cash flow SF CDOs

Deal Name Klio Funding Lakeside II Blue Bell Funding Grenadier

Collateral Rating

At least 85%

‘AA-‘ rated

Min WA Rating

btw AA & AA-

Min WA Rating

btw AA+ & AA

At least 90% 'AA' rated,

avg AA/AA-

Min WACoupon on Fixed Collateral N.A. 5% 6% N.A.

Min WASpread on Floating Collateral 0.70% 0.79% 0.90% 0.55%

Max Fixed-Rate Securities 0% 20% 40% N.A.

Max Floating-Rate Securities 100% 90% 70% 100%

Min Rating 'A-' 'A-' 'A-' 'A-'

Max Rated Less than 'A-' 0% 0% N.A. N.A.

Max Discretionary Trading of Portfolio 10% N.A. 15% 15%

Max WALife (Years) 8 7.5 8 7.5

Portfolio Composition (%) Target Target Target -

Consumer Asset-Backed Securities N.A. 0 0 10%

Commercial Asset-Backed Securities N.A. 0 0 0

HEL, Resi A/B/C 70% (max) 57% 43% -

CMBS N.A. 5% 38% -

Max Pure Private Securities N.A. 5% 10% 0%

Max Payment-in-Kind (PIK) Bonds 5% 5% 3% 5%

Max CDOs 40% 50% 20% 12.5%

Source: Credit Suisse, Fitch, S&P.

Exhibit 25. Examples of high grade cash flow SF CDOs

Deal Name Klio Funding Lakeside II Blue Bell Funding Grenadier

Collateral Manager

Bear Stearns

Asset Management Inc.

Vanderbilt Capital

Advisors, LLC

GMAC Institutional

Advisors LLC ACA Management

Closing Date 4/23/04 3/31/04 12/12/03 7/14/03

Ramp-Up Period (Days) 90 180 90 180

Reinvestment Period (Years) 5 N.A. 5 5

Deal Size 1,263mm 1,502mm 1,250mm 1,500mm

Size of the ABCP Notes 1,074mm 1,170mm 1,112mm 1,320mm

Concentration Limit

Maximum Single Issue (%) 2.0% - -

'AAA' 2.5% - 2.5% 2.0%

'AA-' or Higher 1.5% - 1.5% 1.5%

'A-' or Higher 0.5% - 0.5% 0.5%

Maximum ABS Servicer Concentration (%) 10%. 7.5% 10% 7.5%

‘AA-'/'S1' or Higher N.A. - N.A. 12%

‘A-'/'S2' N.A. - 12.5% N.A.

Below 'A-'/'S2' N.A. - 10% 7.5%

Source: Credit Suisse, Fitch, S&P.

31 March 2006


Why is the size of these deals so large? Currently, the deal size of high grade SF CDOs typically ranges between $700mm and

$1.5bn, which is larger than that of an average CDO. There are several reasons for this.

Investors in senior structured finance paper tend to have larger allocations. This is

because the majority of the investors are banks and large institutional investors who

manage large money market funds, while highly rated tranches or senior classes are often

sizable, allowing for larger allocations. For example, a typical allocation to a senior

investor in a $500mm deal is often in the range of $20~100mm.

In addition, in order to achieve portfolio diversification, CDOs generally apply single-issue

concentration limits. For example, a 2% single-issue limit can be translated to a minimum of 50

bonds per deal. Assuming a $20mm allocation per bond in the pool, this corresponds to a deal

size of $1bn.

Buyers of the senior tranche of these deals (often short-term notes) are mainly money

market investors, who also prefer a larger deal size due to liquidity concerns. Larger deals

tend to have bigger senior classes, which in turn may be distributed to more investors,

resulting in likely better liquidity. In fact, investors in the short-term paper usually prefer

deals that offer $500mm or more in short-term CDO paper. Note all deals in Exhibit 25

issued over $1bn in short-term paper.

Clearly, the relatively tight concentration limits in high grade SF CDOs partly determine the

deal size. Besides the fact that a larger deal may offer greater liquidity, a tighter

concentration limits and/or more portfolio diversification are beneficial to senior CDO

investors from a credit standpoint. Exhibit 25 provides more details for the four recent cash

flow high grade SF CDOs shown in Exhibit 24.

Why do high grade SF CDOs often issue short-term liabilities? These deals often issue short-term obligations, which are often Rule 2a-7 eligible money-

market (MM) tranches. Issuing short-term obligations enhances deal economics as it

reduces liability cost and enhances excess spread and equity return. Because the collateral

for high grade SF CDOs are highly rated, they often carry a lower coupon than that for other

CDOs. To maintain a similar level of excess spread, a lower liability cost is desirable. The

MM structure typically costs less than a term note structure. We illustrate this in Exhibit 26,

where the lower liability cost leads to a 2% additional equity return per annum.

How do the short-term notes work? The short-term obligations may be issued directly from the CDO trust or a separate trust. A

separate trust is sometimes established because if the CDO needs to be a Qualified Special

Purpose Entity (QSPE) it cannot issue a short-term tranche (also known as the MM tranche)

as a QSPE cannot re-issue securities, according to FAS 140. MM tranches must be re-

issued at least every year and the re-issuance usually occurs after some shorter time period

(on the expected maturity date), for example 30 days, 60 days, 90 days or 6 months.

Typically, there are multiple dealers to re-market the notes.

Exhibit 26: MM structure enhances deal economics* Money-Market Structure Term Notes Structure

Put Premium (bps) 20 --

Re-Marketing Cost (bps) 5 --

Class A Coupon (bps) L + 10 L + 45

"All-in" Senior Notes Cost (bps) L + 35 L + 45

Assumed Leverage 20:1 20:1

"Additional" Equity Returns 2% per annum

* Numbers are hypothetical.


31 March 2006


CDOs that issue MM tranches are often structured with an embedded put agreement, from

a highly rated put provider, of which the minimum ratings are usually at least P-1/A-1/F-1

by Moody’s/S&P/Fitch. At the expected maturity date, new MM notes are issued through a

re-marketing (or re-issuance) process, of which the proceeds are used to retire the

outstanding class of MM notes. In the event that the new MM notes cannot be issued at a

coupon less than a pre-set maximum rate,26 the put may be exercised, in which case, a

new class of notes is often issued. This new class of notes carries a pre-set “maximum”

coupon, usually significantly higher than the coupon on the maturing notes. The put

provider’s premium and the pre-set maximum coupon are closely related. A put provider

could potentially charge a lower premium if the maximum coupon is set at a higher limit or

vice versa. Terms are typically dependent on the particular put provider.

Other variations of the short-term notes include extendable notes and medium term notes.

A typical extendable note may have a 13-month maturity and is remarketed each month.

Investors of extendable notes have a monthly option to extend or put the notes. If in any

month investors choose to put the notes, the exercise date would be 12 months later. The

coupon on the notes may step up over time such that a higher coupon would be realized

with continued extension of the notes. A typical medium term note may have a bullet

maturity of two or three years accomplished with a put. It can be remarketed in any

frequency following the maturity of the initial issuance.

What are some key considerations for investors of short-term notes? Typical investors for the MM tranche of high grade SF CDOs include banks and large

institutional investors also securities lenders. Some key investment considerations are as

follows.

1. Investors of short-term notes are credit-enhanced by both structural subordination,

and to some extents by the put provider. Like the senior class for other CDOs, the

credit risk for a MM tranche is largely reduced through structural credit enhancement,

which typically leads to a term triple-A shadow rating for the MM tranche.

Put providers mainly provide liquidity. They are obligated to provide funds (or

liquidity facility) to retire the maturing notes and purchase the newly issued notes.

For this reason, put providers are also referred as liquidity provider. If the put

provider is downgraded below certain rating thresholds without an appropriately

rated replacement counterparty or sufficient posting of collateral, rating agencies

typically require the liquidity facility to be drawn on to maintain the ratings of the

short-term notes.

While the main role of a put provider is to facilitate the re-issuing of the short-term

note, a put provider also credit-enhances the short-term note. For example, even

if the note is downgraded due to collateral performance, it can still carry the same

rating as the put provider so long as the put provider offers sufficient liquidity for

re-issuance. Some deals have certain “out” clauses related to payment default or

event of bankruptcy27, others are related to loss threshold or ratings28. If the “out”

clauses are triggered, the liquidity facility will not be available to the short-term

investors any longer.

26 Or in some cases where the put agreement expires and is not extended, and some other replacement

liquidity is not obtained. 27 The initial put option agreement carries at least the same tenor as the maturity of the MM notes, and in most cases, the put provider is only allowed to terminate the agreement in the event of a payment default on the senior CDO notes or a CDO event of bankruptcy. 28 For example, if the CDO does not have the capacity to make monthly interest payments on the outstanding CP notes, or the realized losses on the portfolio have reached a preset level, the liquidity facility will not be available to the ABCP notes investors any longer.

31 March 2006


Investors of short-term notes are credit-enhanced by both structural subordination

and by the put provider to some extent. However, because of such an

association, MM investors often limit their exposure to a single put provider.

2. Static vs. revolving portfolio

Collateral portfolio can be static or revolving. While static pools are easier to keep

up with, revolving deals with clean guidelines and managed by a reputable

manager can be attractive as managers can play a key role in making collateral

investment decisions. Pools backed by shorter weighted average life assets,

including highly rated real estate ABS, tend to be revolving, i.e., proceeds from

retired collateral can be reinvested in eligible assets to maintain deal leverage.

3. Collateral selection

The average collateral credit rating is high, typically double-A while often no

assets is rated below single-A. For investors who are concerned about some

previously troubled sectors, high grade SF CDOs often apply concentration limits

for synthetic securities, private securities, and PIK bonds, as well as for sectors

such as MH, and other ABS such as equipment, structured settlement and

timeshare etc.

The demand for short-term CDO tranches mainly comes from MM investors, and it has

been healthy. In fact, since the beginning of 2003, money-market tranches have

accounted for about 28% of all SF CDO issuance, mainly because the CDO paper offers a

good spread pick-up over other money-market alternatives such as high-grade corporates

and ABCP. For example, while the CDO short-term paper (with a minimum rating of

P1/A1/F1) can offer L+1~10bps, Tier-1 industrial with 1~3month in maturity currently yields

L-10 to -7bps and Tier-1 US ABCP offers L-4 to –7bps.

In general, high grade SF CDOs are backed by high quality diversified structured finance

pools with an average rating of double-A. The short-term notes issued by these deals offer

attractive spread pickup vs. other money market alternatives, while they are protected by

structural subordination, and to some extent credit enhanced by the put provider. Also, the

liquidity of the short-term notes is enhanced by the put provider and the relatively large

size of these notes.

31 March 2006


A Closer Look at High Grade SF CDOs29

Over the past two years, one of the most significant developments in the CDO market has

been the growing popularity of high grade (HG) SF CDOs. To date, the asset class has

been well received by the market. We think it is essential to examine the sector closely

from both a structural and a collateral perspective, to review some of the recent trends and

address some potential investor concern. We examine 39 HG SF CDOs issued since

2004. The result and methodology used serves as a first-cut screen for investors seeking

opportunities in HG SF CDOs.

Robust Growth in HG SF CDOs The growth in high grade structured finance CDOs has been substantial:

• 43 HG deals priced between January 2004 and early August 2005, totaling

approximately $54 billion.

• Exhibit 27 shows total 2005 YTD issuance of $26 billion, well above the $4 billion

issued in 2002. HG deals represent nearly half of the total SF CDO issuance.

Exhibit 27: Robust Growth in High Grade SF CDOs*

$21 $20

$30$27

$4

$14

$28

$26

15%

41%

48% 49%

$-

$10

$20

$30

$40

$50

$60

$70

2002 2003 2004 2005 YTD

Vintage

Issuan

ce (

$B

N)

0%

10%

20%

30%

40%

50%

60%

% S

hare

of S

F C

DO

s ($

)

High Grade SF CDO ($ BN)

Other SF CDO ($ BN)

% Share by $ Issuance

* Up to mid-August 2005


One reason often cited for the rapid growth in HG SF CDOs is the significant spread

tightening of collateral prevalent in SF CDOs. As the arbitrage embedded in the traditional

mezzanine SF CDO diminishes, CDO issuers are tapping into higher-rated assets, usually

rated Single-A or above. However, we think this is only part of the story. Another driver is,

given the unprecedented growth of the US housing market and the outlook for future

interest rates, some investors are moving up the credit spectrum to achieve the desired

yield through leverage.30

29 This section was originally published in "The CDO Strategist", Issue #7, September 15, 2005.

30 There are two main ways to achieve higher return: moving down the credit spectrum or increasing

leverage.

31 March 2006


The Mechanics of HG Arbitrage How does the arbitrage work for HG CDOs? We address this question from both the

liability and the asset.

Structural features lower liability cost

Senior Funding: Money Market/ABCP Tranche

Most HG deals invest in pools with an average rating of AA, compared to an average

rating of BBB in mezzanine SF CDOs. The high credit quality of the collateral affords a

lower subordination or a larger senior tranche, which is typically 80% to 90% of the total

deal size. By using short-term funding for the senior notes, the funding cost of the CDO

may be dramatically lowered; common approaches include money market (MM) or ABCP

tranches.31

However, to match the asset and liability maturities, the short-term notes are rolled, or

remarketed, as they mature. Should the remarketing not be successful, a put agreement

governed by the ISDA with a highly rated put provider is embedded. When the put is

excised, the notes are put to the put provider, who subsequently owns the term notes with

a step-up coupon, also known as the maximum coupon. The ratings of these classes are

linked to the ratings of the put provider, which are usually F1/P-1/A1. The put provider is

compensated with a premium, usually around 20 bps.

Money-market notes are often issued in combination with medium-term notes (MTN). The

all-in cost for this type of funding is usually around L+35 bps, including the put premium

and the remarketing cost. The all-in cost of issuing ABCP tranches, on the other hand, is

around L+23/24 bps. It seems that ABCP funding has recently gained popularity over

MM/MTN. As shown in Exhibit 28, 22 of the 39 deals in our list use either MM/MTN or

ABCP, and all short-term notes in recent deals are ABCP funded.

Senior Funding: Term Notes

With senior AAA spreads tightening significantly, current AAA spreads stand around L+25

bps (+/-2 bps), slightly higher than the all-in cost of ABCP funding. Unlike ABCP funding,

however, term notes do not have the uncertainty of the ultimate funding costs contingent

upon the success of remarketing and exercise of the put option. As a result, some deals

opt for term funding if the incremental cost over ABCP is only a couple of basis points.

Exhibit 28, shows the split between ABCP and term funding in recent deals at about 50/50.

Interestingly, typically when term funding is used, a delayed-draw note is also utilized to

avoid negative carry during the ramp-up period.

Pro Rata Pay

Many deals in our list also incorporate pro rata amortization schedules in the payment

waterfall provisions. In these transactions, all rated tranches are paid pro rata until the

collateral balance has decreased by half, after which the amortization schedule switches

to the traditional sequential order. The pro rata pay is also subject to all coverage tests – if

any tests fail, the deal will switch to sequential pay.

The reason for the pro rata structure is to pay down the junior tranches with higher costs

faster in order to lower the all-in funding cost.

Lower Subordination and Higher Leverage

As discussed, the subordination requirements for HG deals are lower than mezzanine

deals due to the higher credit quality of the collateral, which also lowers funding costs.

31 Commercial Paper (CP) is any high-quality, negotiable note having an original term to maturity of no

more than 270 days.

Funding Senior

Notes using Money

Market/ABCP

Term funding the

senior notes

Pro rata pay also

lowers the funding

cost

31 March 2006


Collateral sourcing to enhance asset yield

Similar to recent mezzanine SF CDOs, the majority of collateral underlying recent HG SF

CDOs are home equity (HEL; including home equity and subprime mortgages), RMBS and

CDOs. Exhibit 29 shows that, on average, these asset classes account for 52%, 19%,

and 19%, respectively. Exhibit 30 shows the rating breakdown of HG SF CDOs; the

majority is invested in assets rated Aa3 and above.

Exhibit 28: Structural Details for HG SF CDOs (2004-2005 vintages, as of Aug 5, 2005)*

Deal Name Vintage WAR Equity Size Senior Funding

Senior AAA

Spread (bps) Pro-rata Pay

CDO 1 2004 AA+/AA 0.50% ABCP

CDO 2 2004 AA 1.44% Term Notes, Delayed Draw 38

CDO 3 2004 AA 2.37% ABCP yes (50%)

CDO 4 2004 AA+ 0.50% MM and MTN yes (60%)

CDO 5 2004 AA- 1.91% Term Notes, Delayed Draw 37

CDO 6 2004 AA/AA- 1.00% ABCP yes

CDO 7 2004 AA/AA- 1.50% MM and MTN yes (50%)

CDO 8 2004 AA/AA- 1.50% Term Notes, Delayed Draw 35

CDO 9 2004 AA+ 1.55% Term Notes 35

CDO 10 2004 AA 0.63% ABCP

CDO 11 2004 AA- 1.71% ABCP

CDO 12 2004 AA-/A+ 2.30% ABCP

CDO 13 2004 AA/AA- 0.80% ABCP


CDO 15 2004 AA/AA- 1.60% MM and MTN yes (50%)

CDO 16 2004 A+ 1.80% ABCP yes (50%)

CDO 17 2004 AA+ 1.79% Term Notes, Delayed Draw 35

CDO 18 2004 AA/AA- 1.53% Term Notes, Delayed Draw 29

CDO 19 2004 AA 1.59% ABCP yes (50%)

CDO 20 2004 AA- 1.67% ABCP yes (50%)

CDO 21 2004 A+ 2.13% ABCP

CDO 22 2004 2.50% Term Notes 33

CDO 23 2004 AA/AA- 1.50% ABCP yes (50%)

CDO 24 2005 AA+ 1.54% Term Notes, Delayed Draw yes (50%)

CDO 25 2005 AA+ 1.61% ABCP yes (50%)

CDO 26 2005 A 1.49% Term Notes 27 yes (50%)

CDO 27 2005 0.61% ABCP

CDO 28 2005 2.37% Term Notes 32

CDO 29 2005 1.60% ABCP yes (50%)

CDO 30 2005 AA+ 0.61% ABCP

CDO 31 2005 AA+ 1.20% Term Notes 23 yes (50%)

CDO 32 2005 1.54% Term Notes 25 yes (50%)

CDO 33 2005 1.75% Term Notes 25 yes (50%)

CDO 34 2005 AA- 1.60% Term Notes 23

CDO 35 2005 1.84% ABCP yes (50%)

CDO 36 2005 AAA 0.30% ABCP yes (50%)

CDO 37 2005 AA+ 1.34% Term Notes, Delayed Draw 27 yes (50%)


CDO 39 2005 AAA 0.50% ABCP

* The list is sorted by pricing date from the earliest to the latest.

Source: Credit Suisse, Intex, Moody’s, S&P, Fitch

31 March 2006


Exhibit 29: Collateral Distribution for HG SF CDOs (2004-2005 Vintages)

ABS-Auto

0.30%

ABS-Other

1.59%

CDO

18.82%

RMBS

19.27%

ABS-MH

0.25%

CORP

2.92%

CMBS

4.94%

ABS-HEL

51.67%

ABS-Cards

0.25%

Source: Credit Suisse, Intex

Exhibit 30: Rating Distribution for HG SF CDOs (2004-2005 Vintages)

Aa1

7.37%Aa2

26.28%

Aa3

7.31%

A1

5.68%

A2

12.70%Aaa

35.84%

Baa1

0.38%

Baa2

0.15%A3

4.29%


A commonly held opinion is that not all AAA bonds are built alike. There are many ways to

construct a pool of assets with a weighted-average rating of AA. Below is a list of some

methodologies used by HG CDOs to enhance yield.32

1. Buy slow pay AAA home equity or subprime mortgage bonds.

Many AAA bonds in home equity deals are time tranched, i.e., sequentially paid.

The spread difference between a 1st priority AAA bond and a last pay AAA bond

could be up to 25-30 bps, which accounts for the longer average life of a last pay

AAA, normally around six to nine years. The CDO could also invest in AA’s, which

offers about 7-8 bps pick-up (versus last pay AAA’s) and offers a shorter average

life of around 4.5 years.

32 Because of the high leverage, a small increase in the yield of the underlying assets boosts the return of

the equity tranche significantly.

31 March 2006


2. Buy junior AAA CDO tranches.

Most junior AAAs, i.e., non-first-priority AAAs, are currently offered around L+45

bps, about 20 bps wider than senior AAAs. Exhibit 31 lists the rating distributions

of CDO collateral in select HG SF CDOs. Many deals invest a significant portion

in junior AAA CDO tranches.

3. Buy less well-known issuer names.

As an alternative to traditional, more liquid names, HG SF CDOs can pick up

spread by investing a portion of the collateral in less popular or less well-known

issuer names.33 However, currently, strong demand for HEL paper has resulted

in most bonds trading with little to no spread premium, regardless of the issuer

name. At AAA to A levels, the difference is around 5 bps at most. Still, given the

high leverage, even a couple of basis points will make a significant difference: 1

bps translates into approximately 1% under 100-times leverage.

4. Buy seasoned paper.

There may be relative value opportunities in seasoned bonds, and CDOs often

use this approach to pick up extra yield. Exhibit 32 lists the vintage distributions

of the underlying collateral assets of select 2004 vintage HG SF CDOs. Some

deals invest significantly in seasoned bonds, such as CDO 10.

33 Less popular or less well-known names may, but do not necessarily, have less liquidity.

Exhibit 31: CDO Collateral Rating Distribution of Select HG SF CDOs

Deal

Name SNR AAA JNR AAA AA+ AA AA- A+ A A-

CDO 1 55.3% 44.7%

CDO 2 3.9% 34.1% 9.1% 49.0% 1.0% 2.9%

CDO 3 36.9% 44.7% 1.2% 17.2%

CDO 4 42.6% 28.0% 0.7% 25.1% 2.5% 1.1%

CDO 5 24.2% 71.6% 4.3%

CDO 6 100.0%

CDO 7 61.5% 29.8% 4.3% 4.5%

CDO 8 17.7% 29.1% 4.7% 42.2% 6.2%

CDO 10 54.9% 21.7% 17.8% 5.5%

CDO 11 38.2% 41.5% 0.6% 16.6% 2.7% 0.4%

CDO 12 7.5% 70.2% 22.4%

CDO 13 32.6% 13.4% 7.5% 46.5%

CDO 14 5.8% 11.0% 64.2% 16.6% 2.4%

CDO 15 40.4% 33.6% 2.5% 23.5%

CDO 16 55.8% 9.8% 2.2% 22.6% 1.3% 8.3%

CDO 18 4.2% 41.1% 9.0% 41.2% 4.5%

CDO 19 21.8% 26.6% 1.5% 50.1%

CDO 20 17.3% 27.4% 2.5% 52.8%

CDO 21 36.6% 24.6% 35.7% 0.8% 2.2%

CDO 22 25.4% 20.8% 2.4% 49.6% 1.8%

CDO 23 100.0%

CDO 25 27.5% 23.5% 3.1% 45.9%

CDO 28 8.4% 8.4% 32.2% 14.0% 26.6% 10.5%

CDO 30 23.3% 44.8% 4.7% 9.3% 4.7% 13.3%


31 March 2006


The Result

Now let’s examine whether the arbitrage is economic by examining actual asset and

liability spreads. Exhibit 33 shows rough calculations of the aggregate liability cost, WAS,

and expected equity IRR under a zero-default assumption for select deals for which we

have collateral level details.34 We ignore any hedging issues.

Most of the deals have a zero-default IRR from the low- to mid-teens. Some deals have

higher IRR due to the relatively lower credit quality of the underlying pool and thus higher

WAS, such as CDO 12, 16, 21 and 22. However, because of the lower credit quality, the

IRR for these deals will have to be lowered more than the deals with higher credit quality

due to the relatively higher default risk. Some deals have higher IRR simply as a result of

higher leverage, such as CDO 6 and CDO 13.

34 We obtain the collateral information mainly through Intex. Note that there is usually a lag between when a

deal is closed and when it is modeled by Intex. The list is sorted by pricing date.

Exhibit 32: Underlying Collateral Vintage Distribution of Select 2004 HG SF CDOs

Deal

Name 2005 2004 2003 2002 2001 2000 1999 1998

Before

1998 Unknown

CDO 1 4.9% 47.2% 28.7% 3.5% 12.4% 2.9% 0.5%

CDO 2 78.7% 12.6% 2.5% 1.6% 4.7%

CDO 3 6.4% 58.6% 21.1% 2.3% 0.7% 1.1% 2.4% 7.3%

CDO 4 1.8% 60.2% 24.2% 4.9% 3.3% 3.6% 2.0% 0.2%

CDO 5 87.0% 11.2% 1.8%

CDO 6 12.7% 54.2% 16.5% 3.5% 5.4% 0.4% 0.9% 2.1% 3.3% 1.1%

CDO 7 1.9% 49.1% 8.5% 8.9% 3.9% 4.2% 8.8% 1.0% 2.6% 11.1%

CDO 8 5.2% 91.4% 2.3% 0.5% 0.6%

CDO 10 0.8% 75.1% 12.1% 0.3% 9.7% 1.9%

CDO 11 9.4% 71.3% 4.1% 3.1% 1.5% 0.8% 9.7%

CDO 12 7.3% 66.4% 15.5% 6.5% 1.4% 0.1% 1.2% 0.0% 0.8% 0.8%

CDO 13 17.1% 63.7% 9.1% 3.5% 5.5% 1.0%

CDO 14 5.7% 87.9% 3.9% 1.6% 1.0%

CDO 15 0.6% 74.9% 16.1% 5.0% 1.0% 1.2% 1.3%

CDO 16 3.8% 78.8% 9.9% 2.7% 1.9% 1.0% 0.5% 1.3%

CDO 18 6.9% 88.9% 1.9% 0.7% 1.7%

CDO 19 8.2% 81.8% 2.1% 1.9% 2.5% 0.6% 2.9%

CDO 20 14.8% 82.7% 1.6% 0.8%

CDO 21 8.6% 77.6% 9.5% 3.3% 1.0%

CDO 22 7.2% 87.0% 2.7% 0.7% 0.5% 1.8%

CDO 23 9.1% 78.7% 1.6% 0.5% 10.0%


31 March 2006


How much leverage is too much leverage? One of the biggest concerns investors have regarding HG CDOs is the high leverage.

Given the high credit quality of the underlying pool, it is obvious why subordination levels

for HG deals are lower than other CDO types. Exhibit 34 shows the average

subordination levels of select CDO types at different ratings.

However, the question remains: is the subordination level enough to prevent losses to the

notes or is the leverage too high? To answer this question, we have to look at the

historical experience of the underlying collateral. We found Moody’s impairment study to

be one of the more comprehensive sources and we use their results for our study.35

35 Moody's Impairment Rate includes uncured payment default and downgrade to Ca or C.

Exhibit 33: Arbitrage of HG SF CDOs under No-default Scenario

Deal Name WAR Leverage*

Aggregate

Liability Cost

(bps)** WAS (bps)***

Equity IRR

under No

Default****

CDO 1 AA+/AA 200 36.59 57.05 14.92%

CDO 2 AA 70 53.85 82.29 10.73%

CDO 3 AA 42 31.45 75.53 13.11%

CDO 5 AA- 52 51.14 88.64 12.81%

CDO 6 AA/AA- 100 36.37 75.15 25.73%

CDO 7 AA/AA- 67 36.93 72.76 15.22%

CDO 8 AA/AA- 67 42.54 81.46 17.24%

CDO 11 AA- 59 29.37 72.62 17.71%

CDO 12 AA-/A+ 43 36.70 107.01 24.92%

CDO 13 AA/AA- 126 37.49 69.89 24.37%

CDO 14 AA- 67 43.90 83.23 17.71%

CDO 15 AA/AA- 63 33.65 73.01 16.47%

CDO 16 A+ 56 42.21 93.63 21.34%

CDO 18 AA/AA- 65 44.61 83.14 16.68%

CDO 19 AA 63 32.84 72.21 16.63%

CDO 20 AA- 60 31.04 68.86 14.89%

CDO 21 A+ 47 35.18 88.74 19.07%

CDO 22 A+ 40 40.87 99.09 18.09%

CDO 23 AA/AA- 67 35.11 66.44 12.22%

CDO 25 AA+ 62 31.00 69.73 15.94%

* Leverage is calculated as 1 divided by the size of the equity, i.e., 1% equity implies a leverage of 100 times.

** For ABCP tranches, an all-in cost of 24 bps is used. For fixed tranches, an equivalent floating spread is used.

*** Only the floating spread is used.

**** IRR is calculated as (WAS-Liability Cost)*Leverage. A n all-in fee (senior management fee, administration fees, etc.) of 13 bps is

used.


31 March 2006


Exhibit 34: Average Subordination Levels of HG SF CDOs vs. Other CDO Types

T r ip le -B A v e ra g e S u b o rd in a tio n

0 %

2 %

4 %

6 %

8 %

1 0 %

1 2 %

H Y C L O C R E C D O M e z z S F

C D O

H G S F

C D O

S in g le -A A v e r a g e S u b o r d in a tio n

0 %

2 %

4 %

6 %

8 %

1 0 %

1 2 %

1 4 %

1 6 %

1 8 %

H Y C L O C R E C D O M e zz S F

C D O

H G S F

C D O

D o u b le -A A v e r a g e S u bo r d in a t io n

0 %

5 %

1 0 %

1 5 %

2 0 %

2 5 %


C D O

H G S F

C D O

T rip le -A A v e r a g e S u b o r d in a tio n

0 %

5 %

1 0 %

1 5 %

2 0 %

2 5 %

3 0 %


C D O

H G S F

C D O

T r ip le -B A v e ra g e S u b o rd in a tio n

0 %

2 %

4 %

6 %

8 %

1 0 %

1 2 %

H Y C L O C R E C D O M e z z S F

C D O

H G S F

C D O

S in g le -A A v e r a g e S u b o r d in a tio n

0 %

2 %

4 %

6 %

8 %

1 0 %

1 2 %

1 4 %

1 6 %

1 8 %


C D O

H G S F

C D O

D o u b le -A A v e r a g e S u bo r d in a t io n

0 %

5 %

1 0 %

1 5 %

2 0 %

2 5 %


C D O

H G S F

C D O

T rip le -A A v e r a g e S u b o r d in a tio n

0 %

5 %

1 0 %

1 5 %

2 0 %

2 5 %

3 0 %


C D O

H G S F

C D O


Average Subordination Levels of HG SF CDOs vs. Other CDO Types As shown in

Exhibit 35, the impairment rates of A-rated and above are much lower than Baa and below

ratings, except for Aa-rated ABS.36 By applying a constant recovery rate of 55% and

treating impairment as default, we can derive a loss rate matrix based on Exhibit 35, which

is shown in Exhibit 36.

Exhibit 35: Moody's 5-Year Cumulative Impairment Rate by Sector and Original Rating (1993-2004)

Rating RMBS HEL CMBS ABS* CDO ALL Structured Finance

Aaa 1.02% 0.00% 0.00% 0.95% 0.00% 0.62%

Aa 1.45% 0.00% 0.00% 11.64% 1.67% 3.19%

A 1.20% 2.35% 0.66% 2.94% 6.50% 3.23%

Baa 8.45% 6.99% 1.62% 8.43% 25.06% 11.04%

Ba 6.05% 26.88% 3.75% 32.33% 25.56% 16.04%

B 14.93% 41.13% 16.62% 54.31% 53.16% 22.38%

* Exclude Manufactured Housing and HEL.

Source: Moody’s, “Default & Loss Rates of Structured Finance Securities: 1993-2004”, Moody’s Special Comment, July 2005

Exhibit 36: Derived 5-Year Loss Rate by Sector and Original Rating

Rating RMBS HEL CMBS ABS* CDO ALL Structured Finance

Aaa 0.46% 0.00% 0.00% 0.43% 0.00% 0.28%

Aa 0.65% 0.00% 0.00% 5.24% 0.75% 1.44%

A 0.54% 1.06% 0.30% 1.32% 2.93% 1.45%

Baa 3.80% 3.15% 0.73% 3.79% 11.28% 4.97%

Ba 2.72% 12.10% 1.69% 14.55% 11.50% 7.22%

B 6.72% 18.51% 7.48% 24.44% 23.92% 10.07%

* Exclude Manufactured Housing and HEL.

Source: Moody’s, Credit Suisse

36 Here, consistent with Moody's study, "ABS" does not include MH and HEL.

31 March 2006


Unfortunately, Moody’s study does not give us the impairment rates of intermediate ratings,

such as Aa1 or Aa3. To remedy this deficiency, we linearly interpolate the loss rates for

these rating levels based on the rates in Exhibit 36. Exhibit 37 shows the results.

There are several interesting observations that can be made from this matrix:

1. The loss rates of Aa2 and above for HEL and CMBS are zero;

2. Overall, CMBS has the lowest loss rate, followed by HEL and RMBS; and

3. Except for the ABS sector, which excludes HEL and MH, the loss rates at the Aa2

level are significantly lower than the loss rates at Baa2 level.

We apply the loss rates in Exhibit 37 to the underlying collateral of each deal in Exhibit 33

by rating and by sector to calculate an expected loss rate.37 Finally, we divide the equity

size by the loss rate to calculate a loss coverage ratio of the most junior tranche. We

repeat this analysis for each deal with the results shown in Exhibit 38.

We think the overall loss coverage ratio is quite high, as it provides three to four times

coverage for most deals. Certain CDOs offer very high loss coverage, such as CDO 3,

CDO 11 and CDO 7. Although CDO 13 has the lowest coverage ratio at 1.97, the loss

coverage is sufficient since the most junior tranche is rated below BBB-.38 Interestingly,

CDO 13 also has a very high IRR under the zero-default scenario because of its high

leverage, as shown in Exhibit 33. However, there is a CBO tranche with a Moody’s rating

of Baa2 in its collateral pool. This is a situation that investors should examine more

closely. Investors need to pay attention to CDO 14, 16 and 18, as their coverage ratios

are lower than others with similar ratings.

This analysis is a rough estimate requiring further investigation to assess fully the credit

support adequacy of each deal. However, our analysis enables us to take a first look at

each deal in a relatively efficient way. More detailed analysis needs to be conducted on

other tranches as well.

37 We can apply this calculation to all deals if the collateral information is available.

38 The exact rating of this tranche is not available; we only know it is below the BBB- tranche in the capital

structure.

Exhibit 37: Intermediate 5-Year Loss Rate by Linear Interpolation

Rating RMBS HEL CMBS ABS* CDO

Aaa 0.46% 0.00% 0.00% 0.43% 0.00%

Aa1 0.56% 0.00% 0.00% 2.83% 0.38%

Aa2 0.65% 0.00% 0.00% 5.24% 0.75%

Aa3 0.62% 0.35% 0.10% 3.93% 1.48%

A1 0.58% 0.71% 0.20% 2.63% 2.20%

A2 0.54% 1.06% 0.30% 1.32% 2.93%

A3 1.63% 1.75% 0.44% 2.15% 5.71%

Baa1 2.72% 2.45% 0.59% 2.97% 8.49%

Baa2 3.80% 3.15% 0.73% 3.79% 11.28%

Baa3 3.44% 6.13% 1.05% 7.38% 11.35%


31 March 2006


What are the other concerns of HG CDOs? There are some other concerns regarding HG SF CDOs. More specifically, investors are

concerned about the 1) increasing share of the CDO bucket, 2) the synthetic exposure,

and 3) the Single-A and below bucket. We advocate investors watch for these issues and

we discuss each issue below.

1. CDO bucket: Exhibit 35 shows, at the AA and A levels, the impairment rates of

CDOs are generally higher than those of other sectors. Especially at the “A” level,

the impairment rate stands at 6.5%. This probably explains why some HG deals

explicitly restrict any exposure to CDO tranches rated below “Aa3”. We show

actual CDO rating distributions of some select deals in our list in Exhibit 31, and

investors should pay close attention to those with CDO tranches rated below

“Aa3“. Most of the impairments come from the SF CDOs of early vintages, HY

CBOs and IG CBOs. It is also important to assess which CDO sectors are

included in the pool.

2. Synthetic bucket: Generally HG SF CDOs will allow a maximum synthetic bucket

of 25-30%. We suggest investors pay close attention to deals that allow for higher

exposure, especially when it reaches 50%. Synthetic exposure presents certain

unique risks, including: 1) exposure to different counterparties; 2) documentation

risk - mostly differences in credit event definitions; and 3) higher leverage.

3. Single-A or below bucket: As spreads continue tightening, it may become

necessary to go down the credit spectrum in order to achieve the desired yields.

However, the default risk associated with Single-A or below rated assets is

exponentially higher. For example, the impairment rate of A-rated HEL jumps

from zero to 1.06%, as shown in Exhibit 34. In Exhibit 39, we show the rating

distribution of select HG SF CDOs. Investors should identify deals highlighted in

bold to ensure that credit support levels are adequate.

Exhibit 38: Most Junior Tranche Loss Coverage Ratio of Select HG SF CDOs

Deal Name WAR Equity Size

Expected Loss

Rate

Most Junior

Tranche Rating

Loss Coverage

Ratio

CDO 1 AA+/AA 0.50% 0.1861% A- 2.69

CDO 2 AA 1.44% 0.4083% BBB 3.52

CDO 3 AA 2.37% 0.2228% A- 10.64

CDO 5 AA- 1.91% 0.6008% BBB 3.18

CDO 6 AA/AA- 1.00% 0.4003% BBB 2.50

CDO 7 AA/AA- 1.50% 0.2773% BBB 5.41

CDO 8 AA/AA- 1.50% 0.3918% A- 3.84

CDO 11 AA- 1.71% 0.2620% BBB 6.52

CDO 12 AA-/A+ 2.30% 0.6328% A- 3.63

CDO 13 AA/AA- 0.80% 0.4045% 1.97

CDO 14 AA- 1.49% 0.6112% BBB 2.43

CDO 15 AA/AA- 1.60% 0.3771% A- 4.24

CDO 16 A+ 1.80% 0.8035% BBB 2.24

CDO 18 AA/AA- 1.53% 0.6505% BBB 2.35

CDO 19 AA 1.59% 0.4203% BBB- 3.77

CDO 20 AA- 1.67% 0.4522% BBB- 3.69

CDO 21 A+ 2.13% 0.7245% A- 2.94

CDO 22 A+ 2.50% 0.7522% A- 3.32

CDO 23 AA/AA- 1.50% 0.4083% BBB 3.67

CDO 25 AA+ 1.61% 0.3955% BBB 4.08


31 March 2006


Exhibit 39: Rating Distribution of Select HG SF CDOs

Deal

Name Aaa Aa1 Aa2 Aa3 A1 A2 A3 Baa1 Baa2 Baa3 NR

CDO 1 62.60% 5.00% 18.08% 4.88% 3.09% 5.32% 1.03%

CDO 2 45.27% 4.80% 31.70% 6.43% 1.22% 9.55% 1.03%

CDO 3 45.52% 5.83% 32.01% 4.79% 1.69% 8.97% 1.19%

CDO 4 33.68% 1.95% 35.23% 3.97% 4.33% 11.36% 3.24% 6.24%

CDO 5 34.27% 8.01% 28.81% 3.33% 16.55% 9.02%

CDO 6 43.67% 12.78% 19.38% 1.76% 5.25% 13.57% 1.74% 1.45% 0.41%

CDO 7 43.47% 2.40% 24.97% 13.49% 2.66% 12.18% 0.82%

CDO 8 41.89% 5.40% 31.93% 2.06% 2.76% 14.91% 1.03%

CDO 10 37.98% 6.44% 25.94% 9.12% 6.68% 8.01% 5.77% 0.07%

CDO 11 43.01% 17.50% 19.59% 10.53% 3.34% 5.49% 0.54%

CDO 12 33.62% 3.92% 11.02% 3.09% 7.96% 32.03% 8.37%

CDO 13 27.53% 3.21% 52.86% 1.77% 0.38% 13.07% 0.56% 0.62%

CDO 14 27.16% 10.08% 28.20% 9.84% 1.55% 7.99% 15.17%

CDO 15 37.79% 2.64% 28.64% 16.52% 1.40% 7.04% 5.98%

CDO 16 31.53% 0.66% 18.67% 0.30% 3.85% 28.10% 16.90%

CDO 18 40.06% 7.54% 34.68% 2.70% 0.16% 13.02% 1.83%

CDO 19 40.05% 3.56% 39.05% 2.35% 0.49% 11.76% 2.74%

CDO 20 26.55% 1.42% 38.96% 16.61% 0.19% 7.30% 8.96%

CDO 21 25.62% 2.80% 17.07% 1.83% 6.53% 38.30% 7.84%

CDO 22 17.03% 5.89% 12.38% 7.83% 24.19% 23.67% 9.01%

CDO 23 23.10% 9.90% 32.65% 8.97% 19.79% 4.57% 1.00%

CDO 25 35.30% 6.50% 33.66% 1.94% 2.58% 15.21% 4.13% 0.50% 0.19%

CDO 28 1.70% 5.02% 14.39% 7.98% 19.67% 35.77% 9.58% 5.90%

CDO 30 27.73% 5.45% 13.45% 14.28% 8.78% 13.27% 11.67% 3.19% 2.16%


Closing thoughts Overall, we think HG SF CDOs offer a unique risk and return combination. For investors

interested in getting exposure to mortgage-related assets and other structured products

such as CDOs, and have conservative views on the US housing market and interest rate

prospects, HG SF CDOs are a suitable investment strategy – i.e., moving up the credit

spectrum, and using leverage to achieve the desired yield. Although most HG SF CDOs

appear to have sufficient loss coverage, investors should pay close attention to the

individual assets in the collateral, as “the room for error” for HG SF CDOs is relatively

smaller due to high leverage. A single “bad” credit could prove costly, especially to the

lower tranche holders. Our analysis provides a first-cut screen for interested investors.

31 March 2006


High Grade SF CDOs Revisited39

The high grade (HG) SF CDO market continued its vigorous momentum through 2005 and

into 2006 and has surpassed mezzanine SF CDOs by taking more than 50% share of the

SF CDO market. Issuance accelerated in the fourth quarter of 2005: 18 HG SF CDOs

priced totaling $23.65 billion in just one quarter! Given the rapid growth of this market, we

think it is necessary to give our readers an update of this sector.

WARF is trending higher… Data collected on HG SF CDOs over the past three years suggests an upward trend of the

portfolio weighted average rating factor (WARF), one of the most important parameters

measuring the aggregate credit quality of the underlying collateral of a CDO. High grade

SF CDOs, by definition, should have a lower WARF with the average credit quality of the

underlying pool much better than, say, mezzanine SF CDOs. However, as indicated in

Exhibit 40, we think the average credit quality of HG SF CDOs has deteriorated from

AA/AA- to AA-/A+.

Exhibit 40: Change of HG SF CDO WARF Over Time*

0

10

20

30

40

50

60

70

80

90

100

12/10/02 6/28/03 1/14/04 8/1/04 2/17/05 9/5/05 3/24/06

Mo

od

y's

WA

RF

AA

AA-

A+

Source: Credit Suisse, S&P, Moody’s, Fitch, Intex

* Based on all available information obtainable on high grade SF CDOs up through December 2005.

As spreads grind tighter, modest movement down the credit spectrum can certainly help

“juice up” the potential return for equity investors. However, at the same time it also raises

the concern for increased risk of defaults and losses. Therefore, we check to see if there

are changes in the credit support or structure that could mitigate the rise in default risk.

… But, junior subordination declines; leverage rises… The first feature we check is the size of the equity tranche, or the subordination level of the

most junior tranche. Instead of showing just the raw equity size of each deal, which is not

really an apples-to-apples comparison, we want to adjust it using the WARF. Intuitively,

the higher the WARF, the higher the subordination level required – i.e., the bigger the

equity size or the lower the leverage.

39 This section was originally published in "The CDO Strategist", Issue #14, February 16, 2006.

31 March 2006


The way we apply the WARF adjustment is to run a regression – a regression of WARF on

the equity size.40 Next, we calculate the “WARF-adjusted equity size”, or, alternatively put,

the required junior subordination level for a given WARF. Lastly, we take the difference

between the “required” subordination level and the “actual” level and plot it against time.

Exhibit 41: WARF-adjusted Junior Subordination Shortfall/Surplus

-1.50%

-1.00%

-0.50%

0.00%

0.50%

1.00%

1.50%

Oct-03 Jan-04 Apr-04 Aug-04 Nov-04 Feb-05 May-05 Sep-05 Dec-05 Mar-06

Re

qu

ire

d S

ub

ord

ina

tio

n m

inu

s a

ctu

al s

ub

ord

ina

tio

n

Subordination shortfall

Subordination surplus

Deal A

Deal B

Source: Credit Suisse, S&P

As shown in Exhibit 41, a positive number means more subordination should be provided

– i.e., shortfall – and vice versa. How do we interpret the numbers? Let’s use Deal A and

Deal B (marked in the chart) as two examples. First, please note the numbers are only

meaningful in a relative sense. Deal A has a WARF of about 70 (A+) and an equity sized

at 0.8%. However, based on historical experience and the deals in our sample, it

“should“ have an equity sized/junior subordination level of 1.54% instead, a 0.74% shortfall.

On the other hand, Deal B has a WARF of about 43 (AA-) and an equity sized at 1.75%.

Its WARF-adjusted equity size is 1.28% or the “surplus” is 0.47%.41 It seems that over time

the shortfall is climbing: the junior subordination level is declining when adjusted for WARF.

… And no clear improvement in OC test level either… Of course, just the subordination level alone does not provide all the credit support. One of

the other important factors is the tightness of the OC tests. An OC test with a higher

threshold or minimum level, is tighter, and vice versa. A tranche with lower subordination

but tighter OC test could have the same credit protection as a tranche with higher

subordination but looser OC test. Therefore, we check the OC test threshold levels of the

HG SF CDO deals in our sample.

40 I.e., WARF is the X variable, and equity size is the Y variable.

41 The lowest-rated notes of both Bond A and Bond B are BBB-rated.

31 March 2006


Exhibit 42: Most Junior OC Test Threshold Levels

99.8%

100.0%

100.2%

100.4%

100.6%

100.8%

101.0%

101.2%

101.4%

101.6%

Oct-03 Jan-04 Apr-04 Aug-04 Nov-04 Feb-05 May-05 Sep-05 Dec-05 Mar-06

Mo

st

jun

ior

OC

te

st

thre

sh

old

lev

el

Source: Credit Suisse, S&P, Fitch, Intex

Exhibit 42 plots the most junior – mostly BBB or A rated – OC test threshold levels; there

does not seem to be an improvement, but rather, a downward trend,

HG collateral update: share of fixed and CDO assets In terms of collateral composition, we pay special attention to the share of fixed assets and

CDO tranches in the underlying pools.

There is a perception in the market that the share of CDO tranches in HG SF CDOs has

been increasing. Surprisingly, we found the opposite: the percentage of CDO tranches in

the collateral pools actually declined in 2005 (see Exhibit 43), at least based on the deals

with available collateral-level information. Other interesting observations include the rising

share of home equity bonds and the sharp drop of CMBS in the collateral.

Exhibit 43: Asset Allocation of the Underlying Collateral*

0%

10%

20%

30%

40%

50%

60%

2002 2003 2004 2005

Home Equity

CDO

CMBS

RMBS (Resi-A)


* Collateral-level information is not available for all HG SF CDOs in our sample, especially more recent 2005-vintage deals

31 March 2006


Many market participants also believe that the exposure to fixed-rate assets has jumped

as well, and this is a concern because of the convexity risk associated with fixed bonds.

However, we did not find strong empirical evidence supporting this view.

Exhibit 44: Share of Fixed-rate Assets in HG SF CDOs

Total Deals

in Sample*

# of Deals with All

Floating Assets

Fixed %

(all deals)

Fixed % (excl all-

floating deals)

2003-1H 3 0 41.2% 41.2%

2003-2H 5 1 21.8% 27.2%

2004-1H 4 1 20.4% 27.0%

2004-2H 18 9 12.2% 24.0%

2005** 12 3 16.1% 21.2%


* This is the number of deals we have fixed vs. floating information on, not necessarily the same number deals on which other information is

based on

** Not all 2005 deals are included

Exhibit 44 shows the share of fixed-rate assets in HG SF CDOs. There are several notable

points:

1. The high percentage of fixed assets in early 2003 deals is due to significant

exposure to CMBS bonds, such as the Blue Heron deals.

2. In the second half of 2004 in particular, there is a significant number of deals –

50% – with zero exposure to fixed-assets.

3. When all deals are considered, 2005 deals saw an increase of fixed bond share

over the second half of 2004, but overall it is difficult to argue that there is a

significant increase. When the deals with all floating assets are excluded, the

shares of fixed assets are fairly consistent since 2004 – all in the 20’s – and 2005

numbers actually dropped a bit.

Some Comments on the Basis Risk in HG SF CDOs Basis risk is an extremely important issue for HG SF CDOs – every basis point mis-match

could have a significant impact on equity returns given the high leverage ratios. There are

generally four potential sources of basis risk.

1. Difference in payment dates. The payment dates of the underlying assets could

be different from the payment dates of the CDO. As rates fluctuate daily,

especially during volatile periods, there could be a mismatch between the interest

collected from the collateral and the interest paid out to CDO notes.

2. Difference in index rates. Almost all floating home equity bonds, which comprise

the majority of the underlying collateral of recent SF CDOs, pay coupon monthly

and are indexed to 1-month LIBOR. However, the CDO could pay coupon on a

quarterly basis and may be indexed to 3-month LIBOR. As rates of different

maturities do not necessarily move in parallel fashion, the spread between 3-

month LIBOR and 1-month LIBOR could change significantly and cause a basis

mismatch for the CDO.

3. Mismatch between interests collected from fixed-rate assets and coupon

paid out on floating CDO tranches. Hedging vehicles such as interest rate

swaps or caps are typically utilized to mitigate this mismatch, but the challenge is

to match the swap notional with the actual amortization speed of the underlying

collateral, which is difficult to accomplish. Some deals have been granted the

flexibility to revise the hedging notional if and when needed.

31 March 2006


4. Available funds cap risk. The coupon collected from the underlying assets may

not be sufficient to pay interest on the CDO notes due to the embedded available

funds cap risk in home equity bonds. However, as most bonds in HG SF CDOs

are AAA or AA-rated, this risk should be very minimal under normal condition.

In addition to the swap and cap contracts discussed, other techniques such as basis swap

and reserve accounts are used to mitigate basis risks. These certainly help smooth the

mismatch, however, it is very critical for investor to monitor the basis risk of HG deals.

Final comments HG SF CDOs have grown into a mainstay of the CDO market. Due to concerns over the

US housing market, the high credit quality of the underlying collateral in HG SF CDOs is

appealing to many investors and we believe the momentum will continue. That said, given

select unsettling trends we discussed above contained in some of the recent HG deals, we

encourage investors to keep a close eye on these issues.

31 March 2006


Build or Buy: HEL Bonds versus SF CDOs42

We compare two investment strategies: building a portfolio of mezzanine (mostly BBB-

rated) HEL bonds versus buying a BBB tranche of a mezzanine SF CDO containing high

concentration of mezzanine HEL bonds. Based on our analysis of risk/return profiles, we

think the latter investment provides an attractive alternative of investing in the home equity

sector. The main reasons include:

1. Attractive spread pickup: With the majority of mezzanine HEL bonds being

placed in mezzanine SF CDOs, BBB HEL bond spreads are at historical lows of

around L+135/140 bps.43 Yet, SF CDO BBB tranches are still offered at about

L+270 level, representing an attractive spread pickup of about 130 bps.

2. Similar fundamental risks: Given that the majority of the collaterals of recent SF

CDOs are invested in HEL bonds, the risk exposure to the fundamentals of the

home equity sector and housing market are similar in these two approaches,

especially if both invest in the same HEL bonds.

3. More credit support: The BBB tranches of SF CDOs enjoy another layer of

credit protection provided by the equity tranche of the CDOs.44

4. Turbo feature: Most mezzanine SF CDOs have a turbo feature, which allows a

portion of the bond principal to be paid before potential defaults kick in later as

HEL pools season. This feature further enhances the value of the BBB tranche.45

5. Less Available Funds Cap (AFC) risk: SF CDOs should have relatively less

AFC exposure, because the portfolio is more diversified as it could include

RMBS, CMBS, CDOs, or other assets, some of which may not have the same

sensitivity to AFC as home equity bonds.

In our analysis we randomly select a static mezzanine SF CDO priced at the beginning of

2005 and then construct a portfolio ONLY composed of exactly the same HEL bonds,

mirroring those contained in the SF CDO. By applying the identical sets of assumptions

on these HEL bonds, the risk/return profiles of the BBB tranche of the SF CDO and of the

“built” portfolio of HEL bonds are compared. Our baseline prepayment (CPR) and default

(CDR) curves applied to the underlying home equity loans backing these HEL bonds are

shown in Exhibits 48 and 49.46 We then use Intex’s portfolio functionality to generate the

cash flows for each bond as well as the aggregate cash flows of the portfolio.

Exhibit 45 shows the capital structure of the sample SF CDO, whose BBB tranche has a

coupon of L+285 bps. This BBB tranche also has a Turbo feature with a 14% cap on the

equity, i.e., after satisfying the 14% annual return on the equity, the remaining excess

interest will be used to pay down the BBB tranche, subject to passing OC and IC tests.

The collateral of this SF CDO is composed mostly of HEL bonds (Exhibit 46). The majority

of the HEL bonds are rated from Baa1 to Baa3, and are mostly floaters with a weighted

average spread of 227 bps over 1-month LIBOR (Exhibit 47).

42 This section was originally published in "The CDO Strategist", Issue #8, September 30, 2005.

43 Based on our estimates, the percentage of BBB-rated HEL bonds bought by CDOs is around 70-75%.

44 Each rated HEL bond already has the credit support from both the subordination and excess spread.

45 For detailed discussion on the Turbo feature, please refer to "The CDO Strategist - Revisiting Turbo

Structure: Empirical Evidence", July 15, 2005. 46

Since the majority of the underlying loans are hybrid ARM loans, the CPR and CDR curves are more customized to these types of loans.

31 March 2006


Exhibit 50 shows the prepayment, default and severity assumptions for RMBS (Resi-A

mortgages), CDOs, and “others” (includes CMBS, REITS, Credit Card, SBA, etc.). To

simplify the analysis, we use a constant number for each variable and asset type.

Exhibit 46: Rating Distribution of HEL Bonds in the Sample SF CDO

A1A2

A3

Ba1

Baa1

Baa2

Baa3


Exhibit 45: Capital Structure of the Sample SF CDO

Tranche Name Moody's Rating

Percentage of

Total Deal Size Coupon

A1 Aaa 70% L + 32

A2 Aaa 12% L + 50

B Aa2 10% L + 70

C Baa2 4% L + 285

Equity 4%


31 March 2006


Exhibit 47: Baseline CPR Curve of HEL Loans

0%

10%

20%

30%

40%

50%

60%

0 3 6 9 12 15 18 21 24 27 30 33 36 39 42 45 48 51 54 57 60 63

Age (month)

CPR


Exhibit 48: Baseline CDR Curve for HEL Loans

0%

1%

2%

3%

4%

5%

6%

7%

8%

0 3 6 9 12 15 18 21 24 27 30 33 36 39 42 45 48 51 54 57 60 63 66 69 72 75

Age (month)

CDR


Exhibit 49: Assumptions of Other Assets in the SF CDO

Prepayment Rate Default Rate Severity

Baseline

RMBS 20% 0.50% 40%

CDO 0% 0.50% 40%

Other* 0% 0.50% 40%

Stress Level I

RMBS 20% 1.0% 40%

CDO 0% 1.0% 40%

Other* 0% 1.0% 40%

Stress Level II

RMBS 20% 1.50% 40%

CDO 0% 1.50% 40%

Other* 0% 1.50% 40%


* For this CDO in particular, “other” includes CMBS, REITs, credit card, etc.

31 March 2006


As shown in Exhibit 50, the BBB CDO tranche performs better than the HEL portfolio

under the baseline assumption for the rest of the CDO pool and different stress scenarios

for the HEL bonds. The DM holds steady in all scenarios while the DM and IRR of the

portfolio of HEL declines across all scenarios. Notably, when the baseline CDR curve is

doubled, the DM drops by 108 basis points.47 More interestingly, when the forward curve

is bumped up by 200 bps, with a 50% increase in the baseline CDR curve, the DM of the

HEL portfolio drops to 91 bps! Part of the decline is due to the AFC issue, while the AFC

has no impact on the DM of the CDO tranche. In addition, due to the Turbo feature, the

BBB CDO tranche gets paid-down in principal in the first 18 months or so across all

scenarios, which also helps its performance.

One argument to dispute the seemingly stronger performance of the CDO tranche is the

positive benefits contributed by the other asset sectors. However, if non-HEL assets

perform well, this is precisely why we think investors should consider investing in the CDO!

Second, even when stress assumptions on non-HEL sectors are increased to more severe

levels, our main conclusion remains.

In Exhibit 51, we re-run the same analysis on the CDO under different stress scenarios for

the non-HEL sectors, i.e., stressing the CDR to 1% and 1.5% levels.48 At 1% CDR, only in

the scenario where the CDR curve for home equity loans is doubled does the DM of the

BBB CDO tranche starts to drop. However, the magnitude is still smaller than the decline

of the HEL portfolio’s DM in the same scenario. We do not see a similar decline in the

CDO’s DM until we stress the CDR of the non-HEL sectors to 1.5%.49

47 Doubling our baseline CDR curve is a very stressful scenario as the CDO could reach as high as 14%

CDR. However, even under this scenario, the DM of the BBB-rated CDO tranche still hold its DM at the coupon spread level of 285. 48

For CMBS, REITs, and Resi-A mortgages, 1% or 1.5% annual CDRs are very stressful scenarios. 49

Notice that in our analysis, all the prepayment and default assumptions are applied to the underlying loans, such as for RMBS and CMBS, or the underlying bonds, such as for CDOs.

Exhibit 50: Results - Using Baseline Assumptions for Other Assets in the CDO*

BBB Tranche of CDO Portfolio of HEL Bonds

Assumptions Applied on the HEL Bonds DM** IRR DM** IRR

Base Case 285 7.62% 226.6 7.02%

Stress CDR Curve by 50% 285 7.62% 216.5 6.91%

Stress CDR Curve by 100% 285 7.62% 118.9 5.89%

Stress CDR Curve by 50%, and Stress

Forward Curve by 100 bps 285 8.63% 164.3 7.42%

Stress CDR Curve by 50%, and Stress

Forward Curve by 200 bps 285 9.64% 90.9 7.70%

* This analysis is based on the assumption that the CDO will be auction-called at par on the first auction call date and the portfolio of HEL

bonds could also be liquidated on the same date at par.

** For the CDO tranche, the DM is over forward 3-month LIBOR, while for the HEL portfolio, the DM is over forward 1-month LIBOR.


Exhibit 51: Results - Using Baseline Assumptions for Other Assets in the CDO*

BBB CDO at Stress Level I* BBB CDO at Stress Level II**

Assumptions Applied on the HEL Bonds DM IRR DM IRR

Base Case 285 7.62% 285 7.62%

Stress CDR Curve by 50% 285 7.62% 285 7.62%

Stress CDR Curve by 100% 271 7.62% 153 6.28%

Stress CDR Curve by 50%,

and Stress Forward Curve by 100 bps 285 8.63% 285 8.63%

Stress CDR Curve by 50%,

and Stress Forward Curve by 200 bps 285 9.64% 285 9.64%

*Stress Level I increases the CDR of non-HEL collateral to 1%.

** Stress Level II increases the CDR of non-HEL collateral to 1.5%.


31 March 2006


We ran similar analysis on several other CDOs with similar characteristics - recent vintage

(2004 and 2005), mezzanine SF CDOs with large concentrations of HEL, similar structures,

and etc. And while we only showed our results for one CDO, our conclusions hold for all

the deals we have checked. We also believe these results to be consistent across many

of the recent vintage mezzanine SF CDOs with similar characteristics.50 In addition, the

performance of the remaining collateral (i.e. besides HEL) contained in the SF CDOs

needs to be monitored closely to the extent they may impact CDO performance.51

The significant spread pick-up of BBB SF CDOs versus BBB HEL bonds is partially due to

less liquidity and transparency in the CDO space. However, for buy-and-hold investors, a

little less liquidity might be a good source to pick up extra yield. Some might also argue

that the complexity associated with CDO analytics contributes to the premium as well. We

think as investors become familiar with CDOs, this premium will diminish.

The rise of the CDO market has changed the entire landscape of investment. All investors,

seasoned or new, are faced with the challenge of adapting to the new environment. For

investors in the home equity sector, SF CDOs provide an attractive alternative.

Beyond the scope of our analysis, if investors agree that the BBB tranches of recent

mezzanine SF CDOs could offer higher return due to spread pick-up versus a pool of BBB

HEL bonds, yet share similar risk exposure to the housing market and interest rate risk, we

suggest consideration be given to going long the CDO tranche while going short (by

buying protection through credit default swap) some of the individual bonds on which they

have negative views.

50 For interested readers, we would be glad to share more of our results.

51 There could be some extreme situations where the bad performance of certain bonds could hurt the CDO

and make it less attractive.

31 March 2006


Revisiting Turbo Structure: Empirical Evidence

52

The mezzanine turbo feature has become a cornerstone of structured finance (SF) CDO

structures. In 2004, 80% of all managed new-issue SF CDOs included a turbo structure.

Along with the maturation of the CDO market, the turbo structure has evolved too,

incorporating a variety of different approaches.53

Turbo fast-track – how and why they work In a generic turbo structure, the turbo feature serves as a rapid amortization mechanism

for mezzanine tranches typically rated Triple-B. During the reinvestment period (typically

the first 3-4 years for SF CDOs), a portion of the excess interest, after paying liability

coupon payments and fees, and upon satisfying all coverage tests, is used to amortize the

mezzanine notes.54 The methods used to determine turbo amounts vary from deal to deal,

and are explained in detail later.

It is important to note that interest but not principal from the collateral is used to turbo the

notes. This effectively pays down the most expensive tranche, and reduces the liability cost

while avoiding subordination or overcollateralization (OC) erosion of the senior tranches.

What’s the value of, and who benefits from, a turbo structure?

1) For senior notes holders: the senior investors are indifferent.

2) For BBB note holders: the turbo shortens the average life of the bonds. Since

losses in home equity (HEL) collateral do not generally begin until the loans are

about 18 months seasoned, with defaults on BBB HEL bonds usually occurring

even later, the turbo essentially allows a portion of the bond principal to be paid

before defaults, if any, kick in later.

3) For equity holders: the equity holders give up some yield.

A tale of two Turbos Recent deals have included variations to the traditional turbo structure as CDOs have

evolved over time. Exhibit 52 details a few common structures:

Exhibit 52: Common Turbo Structures

Structure Type Duration of Turbo Period

(1) X% Equity cap, Remaining excess interest to turbo BBB Class

(2) Turbo the BBB tranche up to $X per period, Remaining excess interest to Equity

(3) Turbo the BBB tranche based on predetermined schedule, Remaining excess interest to

Equity

(a) For Life of Transaction

(b) During Reinvestment Period

(c) During Specific Time Period

(c) Until Auction Call Date

(d) During Non-Call Period


Additionally, some deals use combinations of the above structures. For example, a deal

may cap the equity at 15% (annual return), divert the excess interest to amortize the BBB

class up to $100,000 per payment period, and then distribute any remaining interest back

to the equity. Furthermore, many deals now-a-days turbo not only the BBB class, but also

other tranches in a reverse sequential order.

52 This section was originally published in "The CDO Strategist", Issue #5, July 15, 2005.

53 In November, 2002, Credit Suisse published a special report titled "Relative Value of Turbo Triple-Bs in

ABS CDOs". 54

For example, OC/IC tests, par value tests, etc.

Turbo defined:

excess interest

diverted to amortize

mezzanine tranches

31 March 2006


To observe the impact of different turbo structures on BBB cash flows, we applied

structures (1) and (2) above on a hypothetical SF CDO deal with characteristics as

specified in Exhibit 53.

Exhibit 53: Sample SF CDO

Deal Structure Deal Information

Tranche Size ($mm) CE(%) Rating Coupon (bps) Reinvestment 3 year Fixed WAC 5.70%

A $308.00 23.0% Aaa L + 30 Turbo Period 3 year Floating WAS 1.75%

B $59.90 8.0% Aa2 L + 58 Auction Call 8 year Floating Assets 76%

C $16.10 4.0% Baa2 L + 265 Total Size $400mm Payment Freq. Quarterly

Equity $16.00 --- --- NA

Base Case Assumptions

Default Rate (Annual) 0.50% Class A/B OC Test 103.5%

Recovery in Default Immediately at 40% Class C OC Test 101%

Class A/B IC Test 115%

Class C IC Test 110%

Turbo Structures

Structure (1) – Equity Cap Structure (2) – BBB Cap

During the Turbo Period, payment to the Equity is capped at 15%

annual rate. Remaining excess interest used to amortize class C

During the Turbo Period, excess interest is first used to amortize

class C, subject to a periodic cap of $162,165. Remaining excess

interest is paid to the Equity


Based on similar recent SF CDO structures, we apply an Equity Cap rate of around 15%.

The BBB dollar cap amount of $162,165 in the second structure is determined by taking

an average of total BBB amortization during the turbo period in the first structure (by

applying a 15% equity cap rate).

Exhibit 54 shows the base case BBB pay-down rate during the turbo period for each turbo

structure. As we can see, the BBB pay-down rate is nearly identical for both structures,

which is as expected given the way we determine the dollar cap amount.

Exhibit 54: Base Case Turbo BBB Pay-down Rate*

0%

2%

4%

6%

8%

10%

12%

14%

0 1 2 3 4 5 6 7 8 9 10 11 12 13Quarters since Closing

BB

B P

ayd

ow

n R

ate

(%

)

0.5% CDR, 15% Equity Cap (Structure 1)

0.5% CDR, $162,165 BBB Dollar Cap (Structure 2)

* The pay-down rate is defined as a percentage of the original BBB tranche balance.


31 March 2006


Exhibit 55: Stressed at 2.05% CDR, Greater Pay-down in Structure 2*

0%

2%

4%

6%

8%

10%

12%

14%

0 1 2 3 4 5 6 7 8 9 10 11 12 13Quarters since Closing

BB

B P

ayd

ow

n R

ate

(%

)

2.05% CDR, 15% Equity Cap (Structure 1)

2.05% CDR, $162,165 BBB Dollar Cap (Structure 2)

* The pay-down rate is defined as a percentage of the original BBB tranche balance.


However, if we increase CDR, the BBB pay-down rates diverge. Exhibit 55 illustrates the effects of CDR increase to 2.05%. It is clear that the pay-down on the BBB is higher in Structure 2 as the same dollar amount ($162,165) implies a lower equivalent cap rate applied on the equity under this default scenario. The equivalent equity cap rate is around 12.94%.55 A more important observation is, should CDR become 2.05% and the equity cap set at 15%, there is a principal loss on the BBB tranche. Under Structure 2 (using the dollar amount cap of $162,165), the par discount margin (DM) on BBB still holds at 265 bps, equal to the coupon spread of this bond. However, under Structure 1 using an equity cap rate of 15%, the par DM turns out to be only 235 bps. To prevent any losses on the BBB tranche, the cap rate has to be lowered to a maximum cap rate of around 12.8%. This illustrates the importance of setting an optimal cap rate to protect the BBB tranche, which is not an easy task.

Historical Performance of Turbo on BBB Tranches How has the turbo structure actually impacted BBB tranches? Using our surveillance universe, we aggregate the actual pay-down experience on turbo BBB SF CDO tranches from 2001-2003, grouped by vintage.56 57 Exhibit 56 shows the results:

Exhibit 56: Historical Turbo BBB Pay-down Experience*

0%

2%

4%

6%

8%

10%

12%

14%

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18

Quarters since Closing

BB

B P

ayd

ow

n R

ate

(%

)

2001 2002 2003

* In this analysis, we exclude CRE (or CMBS) CDOs, real estate CDOs (such as most C-BASS deals), high-grade SF CDOs, and CDO-squared

deals. We also exclude deals with large buckets of CDO tranches (higher than 20%).


55 This time we set the equity cap rate so that the average pay-down is equal to $162,165 under 2.05%

CDR. 56

Please see our Deal Surveillance Reports on www.cbo.com, which tracks the performance CDOs, available to QIB. 57

In this analysis, we don't include CRE (or CMBS) CDOs, real estate CDOs (such as most C-BASS deals), high-grade SF CDOs, and CDO-squared deals. We exclude deals with large bucket of CDO tranches (higher than 20%) as well.

31 March 2006


The chart suggests several interesting points:

• The 2001 vintage experienced the fastest (not necessarily the highest) turbo BBB

pay-down early on - the pay-down rate during the first 4 quarters is much higher

than the other two vintages. We attribute this to slower prepayment speeds of

2001 vintage. In addition, 2001 SF CDOs tend to be comprised of a broader

range of asset classes compared to later vintages, many which prepay slower.

For example, asset classes like corporate bonds (~12% of 2001 pools) and

REITS (~5% of 2001 pools) behave like bullet bonds. In addition, 2001 vintage

HEL loans tend to prepay slower than loans of later vintages. These combined

factors result in slower prepayment and higher principal balance contributing to

more excess spread to turbo the BBB tranche early on.58

• The 2001 vintage also ceased “turbo-ing” the earliest at 1.5 years (6 quarters)

while the 2002 vintage ceased after around two years (8 quarters). In 2002, ABS

downgrades were at historical highs, led by the manufactured housing (MH)

sector. Losses attributed to MH and other troubled ABS sectors during this time

may account for the early termination of the turbo.

• The continued rise of the 2003 vintage turbo BBB pay-down curve may be

attributed to the shift of 2003 collateral pools into more residential mortgage-

related assets, where the credit performance so far has been quite robust.

• From a relative value perspective, the value of the turbo feature in 2001 and 2002

vintage SF CDOs has diminished as deals from these vintages have in general

ceased turbo-ing for at least four payment periods (on a quarterly basis). In

general, the turbo feature in 2003 SF CDOs, however, still has value as the turbo

remains active (for a more detailed discussion on the 2003 vintage of mezzanine

SF CDOs, please see our Strategy section).

Finally, we note that 15 tranches across 7 SF CDOs have been upgraded because of the

turbo structure. Exhibit 57 shows the actions.

Exhibit 57: SF CDOs Upgraded Due to Turbo

Deal Name Class

Orig.

Rating

Prior

Rating

Current

Rating

Rating

Agency Vintage

Capital Guardian ABS CDO I, Ltd. C BBB BBB BBB+ Fitch 2002Q1

E-Trade ABS CDO II, Ltd. B-1 AA AA AA+ Fitch 2003Q3

B-2 AA AA AA+ Fitch 2003Q3

C-1 BBB BBB BBB+ Fitch 2003Q3


Pacific Bay CDO, Ltd. C BBB BBB A- S&P 2003Q4

Equity - - BB+ S&P 2003Q4

Pacific Coast ABS CDO Ltd. C-1 BBB BBB BBB+ Fitch 2001Q3


Pacific Shores ABS CDO Ltd. C BBB BBB A Fitch 2002Q2

PS-CL1 BB- BB- BBB- Fitch 2002Q2

PS-CL2 BB- BB- BBB- Fitch 2002Q2

C BBB BBB BBB+ S&P 2002Q2

Solstice ABS CBO, Ltd. C BBB BBB A- Fitch 2001Q1

St. George CDO Funding 2000-1 Ltd. B AA- AA+ AAA Fitch 2000Q3

C A- A- A Fitch 2000Q3

Source: Credit Suisse, S&P, Fitch

58 Please refer to CSFB periodic report, "Subprime HEAT Update", June 2005.

31 March 2006


Auction Calls in SF CDOs59

The auction call has become a standard feature in nearly every SF CDO. While the legal

final of most SF CDOs is 35 years, reflecting the longest maturity of the underlying

collateral, the auction call mandates that the trustee conduct a collateral auction at some

point after closing (typically 6-8 years in recent deals) to terminate the CDO, effectively

shortening the maturity of the transaction.

The implications of the auction call are significant, especially for equity holders. When a

new deal is being structured and priced, it is usually assumed the deal will terminate on

the first auction call date. However, in reality, whether and when the deal will have a

successful auction could make a difference in terms of the risk/return profile for equity

holders. In particular, in the tight spread and arbitrage environment as equity returns

continue to be squeezed, it is critical that the collateral assumptions are scrutinized and a

sound mechanism is put in place to better predict amortization speeds and loss profile and

ultimately, the probability of a successful auction call.

We discuss auction calls in SF CDOs and establish a framework for assessing the

likelihood of the auction call. The most difficult part of this analysis is that so far there

have been no historical examples of SF CDO auction calls to reference.

Auction Call: How it Works In a typical SF CDO, if any liabilities remain outstanding at the first auction call date (typically

6-8 years after closing), the trustee is required to hold an auction on the underlying collateral.

If the bid prices are insufficient to redeem the outstanding notes plus fees and expenses

and/or meet certain conditions (described in detail below), the auction may be repeated at

each payment date thereafter if holders do not agree to accept less than their redemption

amounts. Simple as it sounds, there are several nuances worth exploring.

The time period until the first auction call date is a function of the mix and amortization

schedule of the collateral. Early vintage SF CDOs (pre-2003) tend to have longer periods

(10 years and up) because of smaller concentrations of residential mortgages and slower

prepayment speeds, whereas CDOs since 2003 have had shorter periods (6-8 years)

because of higher mortgage concentrations and faster prepayments.

The participants of the auction are typically limited to the collateral manager, equity

holders, and the trustees and there must be at least two bidders at any auction (including

the winning bidder) for the auction to be valid. Some deals allow the collateral manager

additional precedence: the manager has the right to purchase the collateral at the highest

bid and/or the right to postpone an auction due to market conditions.

The redemption amount (or purchase price) may also differ from deal to deal, particularly

in the treatment of the equity. Nearly all deals require any and all outstanding notes and

fees and expenses (ex., hedge termination fees) to be redeemed with liquidation proceeds.

For the equity, most deals only require redemption up to the original equity amount, net of

all past distributions. For example, if the original equity amount was $3 million and the

distributions up to the auction date totaled $2 million, the equity redemption amount at

auction would be $1 million. However, if the distributions totaled more than $3 million, the

equity redemption at auction would be $0. In other cases, the equity receives sufficient

payment at auction to achieve a pre-determined annual internal rate of return (IRR).

59 This section was originally published in "The CDO Strategist", Issue #6, July 28, 2005.

Trustee is

responsible for

holding the auction

Who can bid?

Manager, equity

holders, and

trustees

Equity redemption

differs for some

deals

31 March 2006


Furthermore, most CDOs offer the note-holders and/or equity investors the option to

receive less than par, pending a 100% vote from the respective holders.60 In this case, the

decision of whether or not to accept less than par is similar to the call rationale we

established in a past research piece:61

Payment Received From Auction Call Proceeds > Present Value of Future Cash Flows

However, the vote has several caveats. For a deal that is performing as expected, where

the total collateral market value is sufficient to cover all note-holders (including equity) and

fees at auction, the vote has no value as the auction will proceed regardless. However, in

the case of a distressed deal, where the collateral market value provides insufficient

coverage, the vote essentially gives the note-holders and equity-holders more options. By

not accepting the auction call, they can receive more cash flows and a potentially better

total return than settling with less than par redemption. Of course, the risk here is further

losses and deterioration to market value. Depending on the investors’ view of the

collateral and future market conditions, it may be more worthwhile to liquidate the assets.

Finally, to improve the feasibility of collateral liquidation during the auction, in many cases

the collateral manager may divide the collateral pool into a number of sub-pools for

participants to bid on. The composition of the sub-pools is constructed to maximize the

potential sales proceeds. Additionally, this may expand the bid as a single bidder is not

required to purchase the entire pool.

Establishing Prepayment and Default Assumptions We focus on new vintages of SF CDOs, which normally include at least 70-80% of home

equity (HEL) bonds in the collateral pool. It is critical to construct reasonable baseline

prepayment and default curves in order to generate cash flows of underlying HEL bonds

as well as cash flows of the SF CDO.

Nowadays the vast majority of the underlying HEL bonds are floating bonds backed by

mostly hybrid ARM loans, most of which are 2/28 ARMs. As an example, we use a real SF

CDO deal closed at the end of 2004 which has 76% of the collateral in mezzanine HEL

bonds (on average BBB-rated). Exhibit 58 shows the pool profile for the HEL bonds and

general information on our sample SF CDO.62

Exhibit 58: General Information of Assets and Sample SF CDO

Assets

Total Notional 586,063,500 Weighted Average Spread 2.05%

Floating Bonds 97% Weighted Average Coupon 6.03%

Fixed Bonds 3%

Capital Structure of Sample SF CDO*

Share Coupon O/C Test I/C Test

AAA 81.75% L+33

AA 9.75% L+55 104.8% 112%

BBB 4.25% L+310 102.7% 108%

Equity 4.25%

* There is no reinvestment and BBB turbo in this deal.


60 In some cases, a majority vote. The vote is with respect to each class of notes or equity.

61 Please see CSFB CDO Research, "The CDO Strategist - When's the Best Time to Call?", Issue #2, May

31, 2005. 62

The general structure of our sample deal is based on an actual deal, although we simplified the structure slightly. The liability spread levels are consistent with the actual deal, priced in late 2004.

Note-holders/equity

investors can vote

to redeem at less

than par

The collateral pool

could be divided

into sub-pools for

bidding

31 March 2006


Based on CREDIT SUISSE’s proprietary prepayment model we can generate prepayment

(CPR) curves for each bond and we construct an aggregate CPR curve as shown in

Exhibit 59.

Exhibit 59: Baseline CPR Curve of Hybrid ARM HEL Loans*

0%

10%

20%

30%

40%

50%

60%

0 3 6 9 12 15 18 21 24 27 30 33 36 39 42 45 48 51 54 57 60 63

Age (month)

CPR


* A pool mixed with 2/28 and 3/27 ARM loans.

This is a typical CPR curve for hybrid ARM loans with a mix of mostly 2/28 ARMs and

some 3/27 ARMs – there is a spike after 24 months as prepayment usually jumps right

after the rate reset date (2 years for 2/28 hybrids) and there is another spike after 36

months due to 3/27 hybrids for the same reason.

Also, by using CREDIT SUISSE’s proprietary model, we can construct a default (CDR)

curve following a similar approach. Exhibit 60 shows our baseline aggregate CDR curve

for the underlying HEL loans.

Exhibit 60: Baseline CDR Curve for Hybrid ARM HEL Loans

0%

1%

2%

3%

4%

5%

6%

7%

8%

0 3 6 9 12 15 18 21 24 27 30 33 36 39 42 45 48 51 54 57 60 63 66 69 72 75

Age (month)

CDR


31 March 2006


Generating Cash Flows for the SF CDO We assume the entire sample CDO deal is backed only by the HEL bonds that are

contained in the actual CDO deal (42 bonds in total) and apply our baseline CPR and CDR

curves to the underlying loans backing these 42 HEL bonds. We also assume a severity

rate of 40%. Once we have the cash flows of these bonds generated by Intex, we import

these into our CDO model to generate the cash flows.63

To assess whether the deal will be auction-called, we look at several factors: 1) remaining

asset/collateral notional on the auction date; 2) auction redemption amount which includes

the remaining notes balances and the required payment to the equity holders; and 3) the

market value of the remaining collateral on the potential auction call date.64 In order to

have a successful auction, the product of 1) and 3) needs to be no less than 2).

Exhibit 61 shows the factor (current balance divided by original balance) of both the

collateral and liabilities. Normally we expect to see the factor of collateral lie above the

factor of the liabilities. However, under high default/loss scenarios, we could see the

collateral factor dip below the liability factor, which we will discuss later.

Exhibit 61: Collateral and Liability Factors under Baseline Assumptions

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35

Quarter

Factor

Liability Factor Collateral Factor


Both factors are one (i.e., no principal pay-down on both the collateral and liability) during

the first 11 quarters or so because the HEL bonds usually have a 36-month step-down

date, before which all collateral principal payments are distributed to the AAA’s, and the

mezzanine and subordinated bonds normally do not receive any principal.65 Since all of

the bonds in this pool are BBB-rated (some A-rated), there is no principal pay-down early

on. The reason that it is not exactly 3 years is because the bonds bought by the CDO may

already be several months seasoned and the underlying loans may also be several

months seasoned when securitized.

63 We use forward LIBOR curve and convert monthly cash flows of the HEL bonds to quarterly cash flows

for the CDO. 64

In reality, all the expenses, hedging termination fees, and etc are considered. For simplicity, we ignore these. 65

After the step-down date, if triggers are passed, OC is released and a large amount of principal could be channeled to subordinated and mezzanine bonds. However, if triggers are tripped, all principal payment will still go to AAAs and the subordinated and mezzanine bonds receive no principal for as long as the triggers remain tripped.

31 March 2006


How much is the collateral worth on the auction date? It is difficult to forecast the market value of the collateral in 6 or 8 years. Fortunately we

have the projected cash flows, both principal and interest, of the underlying bonds. By

discounting the cash flows after the potential auction date back to the present value as of

the auction date, we can calculate the (theoretical) market price on that date. The question

is: what is the appropriate discount spread to use?

We turn to the ratings transition matrix for an answer. Based on Moody’s ratings transition

on HEL bonds, we can calculate the percentage of HEL bonds, originally rated BBB, that

will migrate to ratings ranging from AAA to CCC in 1 year, 2 years and up to 5 years, if

they are still outstanding. For example, as shown in Exhibit 40, 15.2% of all originally BBB-

rated bonds will be BB-rated in 5 years, if they haven’t been paid off. We further

interpolate the percentage for 6 years, 7 years and up to 10 years, based on which we can

calculate the weighted average rating factor (WARF) and the ratings of the remaining

originally BBB-rated bonds.

Exhibit 62: Rating Transition of BBB-rated HEL Bonds

Actual (Based on Moody’s Transition Matrix) Forecast (Linear Interpolation)

Year 0 1 2 3 4 5 6 7 8 9 10

BBB -> AAA % 0.0% 0.1% 0.3% 0.6% 1.3% 1.8% 2.2% 2.7% 3.2% 3.8% 4.4%

BBB -> AA % 0.0% 0.2% 0.5% 0.8% 1.3% 2.3% 2.6% 3.1% 3.7% 4.4% 5.0%

BBB -> A % 0.0% 1.0% 1.8% 2.6% 3.5% 4.6% 5.5% 6.6% 7.6% 8.8% 10.0%

BBB -> BBB % 100.0% 94.0% 88.0% 79.5% 69.7% 61.3% 52.2% 42.5% 32.3% 21.6% 10.3%

BBB -> BB % 0.0% 2.9% 5.9% 8.9% 12.9% 15.2% 19.1% 22.7% 26.6% 30.6% 34.9%

BBB -> B % 0.0% 0.6% 1.5% 3.1% 5.0% 6.0% 7.7% 9.4% 11.1% 13.0% 14.9%

BBB -> CCC % 0.0% 1.1% 2.0% 4.5% 6.3% 8.8% 10.7% 12.9% 15.3% 17.8% 20.5%

WARF 360 470 567 786 977 1164 1357 1566 1785 2015 2257

Rating BBB BBB/

BBB-

BBB/

BBB-

BBB-

/BB+

BB+/

BB

BB+/

BB BB

BB/

BB- BB- BB-/B+ B+/B

Spread Used (bps) 1200 1400 1600 1800 2000

Source: Credit Suisse, Moody’s

As shown in Exhibit 40, for a BBB-rated pool of HEL bonds, the remaining bonds will

migrate to a BB-rated pool in 6 years.66 The rating will be BB- 8 years later and B/B+ 10

years later. Currently, BB-rated bonds trade between L+850 bps and L+1100 bps area.67

To be conservative, we use the spreads as specified in Exhibit 40.68 69

How much should be paid to the equity holders? In an auction call, the note holders need to be paid the par amount of the remaining

balance. For equity holders, it is a bit more complicated. In most cases, there is either an

explicit or implicit specification that the equity holders be paid back the original equity

amount, net of what has been paid to them before the auction date. In some cases, there

is an IRR hurdle specified so that the amount owed to the equity holders is sufficient to

achieve this IRR.70 Our sample deal uses the first case, which essentially is equivalent to

specifying a 0% IRR.

66 Please notice that we emphasize that it is for the "remaining" bonds, as many bonds have been paid off.

Most of the bonds in our pool have an average life from 3.5 to 4.5 years. 67

For non-Moody's rated BB bonds, the spread is wider and trade around L+1100 bps. 68

B-rated bonds are usually quoted in dollar terms. However, for convenience, we use spread instead. 69

This analysis also ignores collateral changes due to trading in a managed deal. 70

This is equivalent to all the cash flows to the equity holders bring discounted at the IRR and the sum of the PVs has to be equal to the original equity amount.

31 March 2006


Putting it altogether: to call or not to call under the base case Now we are ready to see whether this deal will be successfully auction-called or not, under

our baseline assumptions. Our sample deal has the first auction date 6 years after the

closing date. The equity payment required on the auction date is specified as “the

difference between the original equity amount and all distributions on the equity on any

prior payment date.”

As shown in Exhibit 63, the total market value of the remaining collateral is sufficient to

pay down both the notes and the equity and thus we expect the auction to be successful.

In our sample example, the required payment on equity is simply the difference between

the initial equity amount and total payments paid to the equity holders before the auction

date. It is equivalent to having an IRR hurdle rate of 0%, i.e., discounting all payments at a

discount factor of one. In some deals, the IRR hurdle could be higher. Obviously, the

higher the hurdle, more payment is required and more difficult is the auction.

We used different IRR rates from 0% to 14% to observe how the auction results change.

As shown in Exhibit 64, if the IRR is 10% or higher, the auction will fail on the 1st auction

date. Thus, it is important to pay attention to the magnitude of the IRR hurdle. If an auction

fails on a specific auction date, another auction is held on the next auction date. Under a

10% IRR hurdle rate, the auction is successful in the 26th quarter after closing, as shown in

Exhibit 65.

Exhibit 63: Auction Call Results on 1st

Auction Date (6 years after closing)

Auction Date 6 years (or 24 quarters) after closing

(1) Total Liability Outstanding $100,690,278

(2) Total Collateral Par Value $125,597,977

(3) Total Equity Payment Required* $3,358,177

(4) Market Value at Auction Call Date 90.1%**

(5) Net: (2)*(4)-(1)-(3) $9,152,117

To Call or Not to Call (if (5)>=0, call) Call

* It is calculated as: $24,907,699 (total size of equity) - $21,549,522 (total amount has been paid to equity).

** The pool is expected to be BB-rated and a discount spread over forward LIBOR of 1200 bps is used.


Exhibit 64: Auction Results under Different IRR Hurdle Rates (on 1st

Auction Date)

Equity IRR 0% 2% 4% 6% 8% 10% 12% 14%

Total Liability Outstanding 100,690,278 100,690,278 100,690,278 100,690,278 100,690,278 100,690,278 100,690,278 100,690,278

Total Collateral Par Value 125,597,977 125,597,977 125,597,977 125,597,977 125,597,977 125,597,977 125,597,977 125,597,977

Total Equity Payment 3,358,177 5,037,421 6,994,275 9,265,931 11,894,037 14,925,186 18,411,458 22,411,017

Net 9,152,117 7,472,873 5,516,019 3,244,363 616,257 (2,414,891) (5,901,163) (9,900,722)

To Call or Not to Call Call Call Call Call Call No Call No Call No Call


Exhibit 65: Auction Call Results under 10% IRR Hurdle

Auction Date 24 quarters after closing 26 quarters after closing

(1) Total Liability Outstanding $100,690,278 $76,996,618

(2) Total Collateral Par Value $125,597,977 $101,904,317

(3) Total Equity Payment Required $14,925,186 $14,767,534

(4) Market Value at Auction Call Date 90.1% 91.8%

(5) Net: (2)*(4)-(1)-(3) ($2,414,891) $1,808,778

To Call or Not to Call (if (5)>=0, call) No Call Call


31 March 2006


What if prepayments are slower? In a slower prepayment environment, the pay-down speeds on both asset/collateral and

liability sides will slow down.

Exhibit 66: Collateral & Liability Factors under Slower Prepayment Assumption*

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35

Quarter

Factor

Liability Factor_Slower Prepayment Collateral Factor_Slower Prepayment

Liability Factor_Baseline Collateral Factor_Baseline


* We stress the baseline prepayment (CPR) curve by 25%, i.e., a previous 10 CPR will be stressed to be 7.5 CPR.

As shown in Exhibit 66, in the 32nd quarter, the collateral factor under the slower prepayment

scenario is 23% versus 7% under the baseline assumptions; while the liability factor under

the slower prepayment scenario is 20% vs. 4% under the baseline assumptions.

Exhibit 67: Auction Call Results on 1st

Auction Date under Slower Prepayment

Auction Date 6 years (or 24 quarters) after closing



(3) Total Equity Payment Required $0

(4) Market Value at Auction Call Date 79.6%

(5) Net: (2)*(4)-(1)-(3) ($17,597,793)

To Call or Not to Call (if (5)>=0, call) No Call


There are a couple of interesting observations. First, note that the required payment on

equity is zero. This is because under a slower prepayment speed, more interest is

generated from the collateral, and the equity is paid faster to the extent that the original

amount is all paid off by the first auction date. Second, the projected market value on the

auction date dropped significantly - from 90.1% in the baseline scenario to 79.6%. This is

because the outstanding balance of the underlying collateral is much higher due to slower

prepayment and thus larger denominator.71 It might make economic sense also: under our

set-up, the originally BBB-rated pool migrates to a BB-rated pool; if this BB-rated pool has

higher balance, it means relatively more bonds “go bad”. Under slower prepayment the

remaining BB-rated pool is larger and if it can be regarded as a worse credit performance,

it should get reflected in the market price.

71 Even though the numerator is also higher as cash flows are more back-end loaded, i.e., more cash flows

come after the auction date.

31 March 2006


As it turns out, under slower prepayment scenario, this deal will not have a successful

auction until the 34th quarter after closing. However, in reality, the note-holders have the

right to vote to receive less than par for the auction to succeed.

More importantly, in this case, the equity holders can actually receive more cash flows if

the deal is not called while receive nothing from the call proceeds if called.72 However,

since the equity holders have already received 100% of the original amount, they cannot

stop the auction call.

What if default rate is higher? – Call fails in distressed scenario In a distressed scenario, the auction call usually fails. As an illustration, we stress our

baseline default (CDR) curve by 250% (or two and a half times). Exhibit 68 shows the

collateral and liability factors and, as expected, both factors decline very slowly and the

collateral factor actually drops below the liability factor at around the 25th quarter. Exhibit

69 also shows the cumulative loss of the underlying pool of HEL bonds and the junior OC

test. The junior OC (BBB OC) test failed around 20th quarter.

As expected, the auction call fails on the first auction date. Exhibit 70 shows even on the

40th quarter, the auction call still fails.73

Exhibit 68: Collateral and Liability Factors under Higher Default Assumption*

30%

40%

50%

60%

70%

80%

90%

100%

1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35

Quarter

Factor

Liability Factor Collateral Factor


* We stress the default rate (CDR) by 250%, i.e., an original 5 CDR will be stressed to 12.5 CDR.

72 The equity holders will receive any surplus of the call proceeds after paying all the fees and note holders.

However, they could face a situation as follows: suppose a bidder views the true value of the collateral to be higher than 79.6% and bids where the proceeds are just enough to pay note-holders while nothing is left for equity holders. The equity holders can’t do anything to stop it. If the equity holders also believe the true value of the collateral is higher, they should try to win the bid at a price they believe to be lower than the liquidation value of the collateral. 73

Under this stressed scenario, the senior OC level drops below 100% in the 36th quarter. Technically, this CDO is then in default and should be liquidated, however, this is a separate topic.

31 March 2006


Exhibit 69: Cumulative Loss of Underlying Collateral and Junior OC Test

0%

2%

4%

6%

8%

10%

12%

14%

16%

18%

20%

1 4 7 10 13 16 19 22 25 28 31 34 37 40 43 46 49 52 55 58 61

Quarter

Cu

mu

lati

ve

Lo

ss

75%

80%

85%

90%

95%

100%

105%

110%

1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39

Quarter

OC

Te

st

Junior OC Target Junior (BBB) OC Level


Closing thoughts In general we can break down the auction call analysis into three scenarios:

1. If the deal performs very well, it is most likely that both the note-holders and

equity holders will be paid 100% and the auction call is automatically executed.

2. If the deal performs very poorly, it is likely it will not have a successful auction call.

3. If the deal performs “in between”, the auction might be successful but not

necessarily on the first auction date. The note-holders and equity-holders might

be willing to receive less than par if they believe the PV of future cash flows is

less than that from the auction call proceeds. However, how much in received

proceeds depends on the market value of the collateral and how much bidders

are willing to pay. This is the beauty of the auction process yet makes the

analysis difficult.

Most important, having reasonable prepayment and default assumptions is crucial for

auction call analysis of new SF CDOs with a vast exposure to HEL bonds. We provide a

framework, rather than a conclusion, for analyzing the auction call.

Unfortunately, there is no empirical evidence, as no SF CDO has been auction called yet.

It will be useful and interesting to observe different constituents’ behaviours in an auction.

Exhibit 70: Auction Call Results on the 40th

Quarter under Stressed Scenario

Auction Date 40 quarters after closing



(3) Total Equity Payment Required $10,535,428

(4) Market Value at Auction Call Date 68.6%*

(5) Net: (2)*(4)-(1)-(3) ($100,106,313)

To Call or Not to Call (if (5)>=0, call) No Call

* We still use the same spreads as in the baseline scenario. Arguably, given the stress level of this scenario, the market value of collateral

could be stressed even lower.


31 March 2006


Default Assumptions for BBB HEQ in SF CDOs

74

Executive summary In this report, we will answer the following two questions: 1) How do we calculate HEQ

default rate in order to fit the special needs of CDO modeling? and 2) What is the

reasonable baseline default rate assumption for triple-B HEQ bonds?

We think Moody’s impairment rate is a better approximation for the default definition in

typical recent SF CDOs.

Based on how CDO pools are typically constructed, we also suggest using an “Aging

Curve” to calculate default rate, instead of “Cohort Rate” or “Lifetime Rate.”

1. Based on historical experience of impaired HEQ bonds, we suggest using a step

curve of impairment/default rates, or a vector of rates, to represent the baseline

annual default rate for triple-B HEQ bonds for SF CDO modeling.

2. By comparing the expected return on equity, we also conclude that using this

vector of rates is equivalent to using a flat annual rate (CADR) of 75 bps.

We hope our analysis will inspire further research. Over time, as more empirical evidence

accumulates, additional insight will be revealed on this topic.

Overview It is important to use reasonable baseline default assumptions for mezzanine HEQ bonds

in SF CDO modeling, as most mezzanine75 SF CDOs today may contain about 50% HEQ

bonds.

However, arriving at reasonable default assumptions for HEQ bonds is not a simple

process. First, there is an inconsistency between the definition of default used in most

default studies and that used in CDO documents. Secondly, commonly used default

statistics, such as cohort and lifetime rates, are inappropriate for SF CDO modeling.

With little empirical evidence on defaults for structured finance bonds, the market has

looked at corporate bond experience for guidance. There are two main approaches. The

first used by some market participants, including Fitch and S&P, begin with a commonly

held view that “diverse” portfolios containing ABS and MBS should have a lower default

risk profile than one with corporate debt securities. They adjust the default matrix of

corporate bonds to reach a default matrix for SF bonds. 76 Alternatively, the second

approach (used by Moody’s) makes use of the “Idealized Loss Rate” table both for

corporate bonds/loans and for SF bonds. By assuming a recovery rate, the default rates of

SF securities could be “backed out” and applied to CDO modeling. 77 A common

weakness is little or no empirical evidence to support these approaches.

74 This section was originally published in "The CDO Strategist", Issue #1, May 11, 2005.

75 Typical mezzanine SF CDOs have WARF around 360 (BBB level), while high grade SF CDOs usually

have WARF around 20 (AA level). 76

See Fitch's special report titled "Rating Criteria for Cash Flow ABS/MBS CDOs", November 9, 2000, and S&P’s special report titled “Global Cash Flow and Synthetic CDO Criteria”, March 21, 2002. 77

Moody’s also apply different stress factors to the default rate used depending on target rating. For example, for a tranche to be rated Baa2, a factor of 1.23 will be used.

31 March 2006


A definition of default tailored for SF CDOs For CDOs, the definition of “default” for the underlying bonds ultimately affects the

performance of the CDO. Once an underlying bond is designated as “defaulted,” it will be

marked down to the lesser of market value OR estimated recovery value for the purpose

of over-collateralization (OC) calculation. As the calculated OC is reduced, it may trigger

pay-down of senior notes. A broader definition of default can thus lead to more and/or

earlier pay-down, which potentially affects the performance of both debt and equity.

In our opinion, we think an “impairment rate” might be a more suitable default

measurement for SF CDO modeling.

In a CDO, a structured finance security (SFS) is considered in “default” under the following events:

1. There is a principal write-down or a PIKable bond that has not received interest payments for typically more than 6 months, or,

2. A rating downgrade to Ca/C or lower by Moody’s, CC or lower by S&P or Fitch.

According to Moody’s, an impaired security is one that has:78

1. Sustained payment default (including principal write-down and interest shortfall) that has not been cured, or,

2. Downgraded to Ca or C (and is therefore expected to suffer a significant level of

payment losses in the future)

While we believe the impairment rate is a more suitable default metric for SF CDOs, it is also a broader definition of default and therefore more punitive. It should be noted that it is possible for deferred cash flows owed to an impaired security to be eventually paid off in their entirety, avoiding an ultimate loss. However, it is often classified as a defaulted security by a CDO should it continue to defer payments for a period of time. Similarly, a bond that is downgraded to double-C may not incur an ultimate loss, but is also typically treated as a defaulted security in CDOs.

Life-time rate & cohort rate: not suitable for CDO modeling How do we calculate a suitable default rate? First, let’s take a look at two popular approaches used by rating agencies: “Cohort Rate” and “Lifetime Default Rate”:

1. “Cohort Rate” – A cohort includes all outstanding HEQ bonds issued up to (and including) the beginning of the cohort unit: January 1 for annual cohort, or the first day of the month for monthly cohort. Should a bond default, it is included in the rating bucket that it started with at the beginning of the cohort. For example, take a bond issued in December 2000. Suppose it’s original rating is “BBB,” and it defaults in January 2003, the bond is included in the calculation of two-year (from January 2001 to January 2003), BBB default rate for the 2001 cohort.

2. “Lifetime Rate” - Lifetime rates generally include all bonds issued during a specific timeframe.

A simplified example shown in Exhibit 71 clarifies the differences among the three default definitions: cohort rate, lifetime rate and vintage rate.79 Suppose bonds A1 and A2 are issued in January, 1999, B1 and B2 in January, 2000, and C1 and C2 in January 2001, etc.. The 2002 Cohort has eight bonds from A1 to D2; the 2002 Vintage has two bonds

78 “Payment defaults and material impairments of U.S. structured finance securities: 1993~2003,”

September 2004, Moody’s. 79

The vintage rate here is a cumulative rate. For example, 2001 vintage rate is the total number of bonds defaulted divided by total number of bonds in 2001 Vintage, as of a specific date.

“Impairment Rate”:

A Better Fit for

CDOs

Impairment seems

to fit CDOs’ default

definition

Impairment is often

more punitive than

default

31 March 2006


(D1 and D2); and the 2002-2005 Lifetime (lifetime rate always refers to a specific time period) includes 8 bonds from D1 to G2. For simplicity, we’ll assume that bonds A1, A2, B1, B2, C1, D1, and E1 all default in Jan. 2005. There are more defaults in older bonds as default risk increases with seasoning.

Exhibit 71: Comparing Differences: Cohort, Lifetime and Vintage Rates

Jan, 1999 Jan, 2000 Jan, 2001 Jan, 2002 Jan, 2003 Jan, 2004 Jan, 2005

A1, A2 B1, B2 C1, C2 D1, D2 E1, E2 F1, F2 G1, G2

2002 Cohort: A1, A2, B1, B2, C1, C2, D1, D2

2002 Vintage: D1, D2

2002 - 2005 Lifetime: D1, D2, E1, E2, F1, F2, G1, G2

Bonds defaulted in Jan, 2005: A1, A2, B1, B2, C1, D1, E1

2002 Cohort Default Rate (as of Jan, 2005): 6/8 = 75%

2002 Vintage Default Rate (as of Jan, 2005): 1/2 = 50%

2002-2005 Lifetime Default Rate: 2/8 = 25%


We think the “Cohort Rate” and the “Lifetime Rate” are unsuitable for CDO modeling. The

primary reason is that CDOs typically invest mostly in newly-issued bonds.80 As a result,

most bonds in one CDO are usually in the same vintage as of the effective date.

1. Cohort rates tend to be too high (too conservative). Since cohort rates

include seasoned HEQ bonds from prior years and because defaults tend to rise

as loans season, the cohort rate is usually higher than a vintage rate. As shown

in Exhibit 71, the 2002 vintage rate is 50%, while the 2002 cohort rate is 75%.

2. Lifetime rates tend to be too low (too optimistic). When calculating the lifetime

rate, bonds defaulted during the period are divided by all the bonds issued in the

same period which include some bonds issued during the back-end of the period

(i.e., less seasoned). Because newer bonds have lower default risk, the inclusion of

less-seasoned bonds will inherently “dilute” the true default rate. The example in

Exhibit 71 shows that the 2002-2005 lifetime rate is 25%, much lower than the

2002 vintage rate as of Jan, 2005. As HEQ issuance has grown rapidly in recent

years, the bias could become more significant if lifetime rate were used.

Our solution: aging curve of impairment rates Ideally, for CDO modeling purposes we should use a default curve that reflects bond

seasoning. Given the small sample of impaired bonds – a total of 17 bonds in our sample

– we create one aggregate aging curve instead of different aging curves by vintages. Here

is how we derive the aging curve of impairment rate:81

1. Line up all HEQ bonds originally rated BBB by age (month) and count the number of

bonds at each age, and mark those bonds deemed “impaired” (Sub-Appendix I), 82 83

2. Calculate the impairment rate by dividing the number of bonds impaired by the

total number of bonds at each age (month).

80 For static deals, collaterals are generally accumulated within 3~6 months before/after the closing date;

For managed deals, limited discretionary or credit trading is permitted, which may result in purchasing some seasoned bonds. 81

This approach has actually been widely used in cash flow modeling for most ABS such as home equity deals: CPR or CDR curves by age are usually created to generate cash flows. 82

Our default rate is based on deal count (not outstanding balance). This more closely reflects the nature of CDO pools, to the extent CDO collateral guidelines impose concentration limit on single-name exposure. 83

The list in Sub-Appendix I excludes corporate guaranteed or wrapped bonds.

Cohort and Lifetime

Rates: Caveats

31 March 2006


Exhibit 72: BBB HEQ Bonds Impairment Rate by Age (Month)

We include HEQ bonds initially rated BBB (BBB+/Baa1, BBB/Baa2, BBB-/Baa3) and also use the lowest rating across three

agencies.

0.0%

0.5%

1.0%

1.5%

2.0%

2.5%

3.0%

0 6 12 18 24 30 36 42 48 54 60 66 72 78

Age of Bond (Tranche)

Imp

air

men

t R

ate

Source: Credit Suisse, Moody's, S&P, Fitch, Intex

Exhibit 72 shows the result and there are several key points:

1. There are usually no impairments prior to Month 36 (in our sample, only two

bonds were impaired between Month 30 and Month 36). This is largely due to the

step-down date of a HEQ deal which is usually 36 months.84 Also, Exhibit 73

shows the cumulative loss of the underlying loans typically stays very low early

on, and usually does not accelerate until approximately Month 18. It also appears

that for a bond to be exposed to impairment risk – usually after Month 36 - the

cumulative loss will reach at least 2%, as shown by Point A in Exhibit 73.

2. The curve in Exhibit 72 is highly non-linear, which suggests that linear

interpolation or taking a simple average of cumulative default rates is

questionable.

3. Since we track all BBB-rated HEQ bonds from 1998 to March 2005 in our

database and Intex, we believe that this is a comprehensive and collective result,

capturing the entire “cycle” of the HEQ sector thus far.

84 The step-down date is a very important date in typical HEQ deals. Before the step-down date, all

collateral principal payments are distributed to AAAs, the mezzanine and subordinated bonds normally do not receive any principal. After the step-down date, if triggers are passed, OC is released and a large amount of principal could be channeled to subordinated and mezzanine bonds. However, if triggers are tripped, all principal payment will still go to AAAs and the subordinated and mezzanine bonds receive no principal for as long as the triggers remain tripped.

31 March 2006


Exhibit 73: Cumulative Loss of HEQ Loans by Vintage

Fixed and ARM Combined

0%

1%

2%

3%

4%

5%

6%

0 6 12 18 24 30 36 42 48 54 60 66 72 78 84

Age

Cu

m L

oss

1998 1999 2000

2001 2002 20032002

2003

Source: Credit Suisse, "Subprime HEAT Update", February 2005.

Determining the impairment rate at bond level based on the cumulative loss experience of

the underlying loans could be challenging. Because credit protection and subordinated

classes act as barriers for senior classes, loan level loss experience may differ

significantly from the loss experience of the securities supported.

The rates in Exhibit 72 are not annualized. To derive an annual impairment rate, we count

the number of bonds impaired in the past 12 months at each age (month), and divide this

by the total number of bonds outstanding 12 months ago. We continue the calculation by

rolling one month each time. For example, at age of 40 (months), we have 2 bonds

impaired in the last 12 months, and with a total of 553 bonds outstanding 12 months ago

(at Month 28), we end up with a 12-month impairment rate of 0.36% (2/553). One month

later, at age 41, because there is one more bond impaired in the 41st month, we have 3

bonds impaired with 519 bonds outstanding at Month 29, resulting in a 12-month rate of

0.58%. The Curve A in Exhibit 74 is the 12-month rolling impairment rate and the Curve B

is a step function we fitted to Curve A. The functional form of Curve B is:

We recommend using this step curve of impairment rate for SF CDO modeling. As BBB

HEQ bonds usually do not become impaired in the first 36 months, we think a zero

impairment rate for the first 36 months reflects the reality. While the rate appears high at

the back-end of the curve, one has to keep in mind that typical BBB HEQ bonds have an

average life of 3.5-5 years. This fact has two implications: 1) bonds with the majority of

their balance still unpaid after 66 months are more likely to be in trouble;85 and 2) as most

of the bonds do not become impaired and the majority of their balances are paid off during

the first 5 years, more weights are placed on the earlier years.

85 For example, if a deal has been performing badly and, as a result, triggers are tripped and OC can not be

released to pay down subordinated and mezzanine bonds after step-down date, the balance of BBB bonds could remain unpaid for a while until the triggers are passed.

66,650

6648,230

4836,40

360,0

)(

><=<<=<<=<=

��

�

�

=

ageif

ageif

ageif

ageif

bpsRatempairmentI

A

31 March 2006


Exhibit 74: Annual Impairment Rate vs. Fitted Step Curve of Impairment Rate

Curve B: Fitted Step

Curve of Impairment

Rate

Curve A: 12-Month

Rolling Impairment

Rate

0

100

200

300

400

500

600

700

800

900

0 6 12 18 24 30 36 42 48 54 60 66 72 78

Age of Bonds

Imp

air

men

t R

ate

(b

ps)


The Impact on Equity Returns We used a generic cash flow model of a mezzanine SF CDO to illustrate the impact of our

vector of BBB HEQ default rate – the step function - on the expected equity returns

(Exhibit 75). Current market spreads for both assets and liabilities were used in our model

and the portfolio contains: 50% HEQ, 20% Residential-A mortgages, 15% CBO tranches,

11% CMBS, and 4% other ABS assets such as credit card receivables.86 We changed the

baseline default assumptions for HEQ bonds while holding the default assumptions

constant for the remaining collateral to see the sensitivities of expected equity returns to

different default assumptions of HEQ bonds (Exhibit 75).

Using our aging curve of impairment rate, our model generates an expected return of

13.45% for equity. We also used three (annual) flat impairment rates – 75 bps, 60 bps

and 25 bps. The expected return using 75 bps is very close to the result of using the step

curve. Thus, if a flat default rate is needed, we would suggest using a constant annual

default rate of 75 bps for BBB HEQ bonds.

86 We used a constant annual default rate for Residential A, CBO, CMBS and other ABS, which are 10, 20,

25 and 35 bps respectively.

Exhibit 75: Default Assumption of BBB HEQ Bonds vs. Expected Equity Return

Scenario

Baseline Annual Default Rate ofHEQ

BBB Bonds (bps)

Aggregate Default

Rate (bps)

Expected Equity

Return

Using our Step Curve A vector of impairment rates in Equation 1 NA 13.45%

Using a Flat Impairment Rate 75 50 13.46%




31 March 2006


Appendix I. Impaired BBB* Rated HEQ Bonds (1997~2004)

Appendix II. Cumulative Impairment Rates by Vintage

Rather than calculating aging curves by vintages, we calculate in Exhibit 2 cumulative and

annual impairment rates of each vintage to show the different performance of different

vintages: 1998 vintage has the highest annual rate of 1.54% followed by 2000 and 1999

vintages with 1.21% and 0.87% respectively; 2001 vintage come in the middle of the ranking

with 0.42%; newer vintages after 2002 all have zero impairment rate so far.

There are two problems in using the annual rate in Exhibit 77:

1. As the performance is dramatically different across vintages, we face the

dilemma of picking which vintage(s) to use. Trying to assess vintage idiosyncratic

behavior is not an easy task and may not be reliable. Please refer to Appendix II

for a description of the evolution of HEQ market.

2. Because the impairment curve by age is non-linear, as we pointed earlier, taking

a simple average of a cumulative rate is questionable.

Exhibit 76:

Issuer # of bonds Vintage

ContiMortgage Home Equity Loan Trust 13 1997, 1998, 1999

GE Capital Management Services 4 1996, 1997, 1998

Delta 4 1997, 2000, 2001

IMC Home Equity Loan Trust 3 1997, 1998

IndyMac Home Equity Mortgage Loan 3 2000, 2001

Conseco Finance Home Equity Loan 2000-B 1 2000

AMRESCO Residential Mortgage Loan Trust 1998-1 1 1998

Ocwen Residential MBS Corp. Mortgage Pass-Through, 1998-R3 1 1998

Southern Pacific 1 1997

Saxon 1 2000

The list excludes corporate guaranteed or wrapped bonds. * This include HEQ bonds initially rated BBB (BBB+/Baa1, BBB/Baa2, BBB-/Baa3).

We also use the lowest rating across three agencies

Source: Credit Suisse, Moody’s, S&P and Fitch.

Exhibit 77: BBB* HEQ Cumulative Impairments by Vintage

HEQ

Issuance year

Total # of impaired

original rated BBB HEQ

issued in the year(1)

Total # of original

rated BBB HEQ

issued in the year(2)

Cumulative%(3)=(1)

/(2)

Annualized%(4)=(3)/

(years to date)

1998 8 74 10.8% 1.54%

1999 3** 54 5.5% 0.87%

2000 4 70 5.7% 1.21%

2001 2 144 1.4% 0.42%

2002 0 317 0% 0.00%

2003 0 697 0% 0.00%

2004 0 1122 0% 0.00%

* This include HEQ bonds initially rated BBB (BBB+/Baa1, BBB/Baa2, BBB-/Baa3). We also use the lowest rating across three agencies. ** All

three impairments are ContiMortgage bonds.

Source: Credit Suisse, Moody’s, S&P, Fitch

31 March 2006


Appendix III. The Evolution of HEQ Market Pre-1996 Nascency

In the early 1990s, home equity loans generally referred to second-lien loans. The market

was dominated by a few specialized lenders, and mainstream banks active in mortgage

financing did not actively participate.

1996~1998 Initial Growth

The HEQ market grew rapidly between 1996 and 1998, almost tripling in annual issuance.

This is largely attributable to: 1) A few mainstream banks began to lend to the subprime

market; 2) More non-bank lenders were able to make loans as securitization offered an

alternate source of funding; 3) Improved credit scoring technology helped lenders to better

understand and price borrower credit risk. During this period, the market shifted from

second-liens towards subprime first-liens. Also, strong competitive pressures led to the

loosening of underwriting standards and the introduction of higher LTV programs by new

lenders to capture market share.

1999~2001 Consolidation

There was a major shakeup among subprime lenders between 1999 and 2001. Three-

fourths of the active lenders either exited due to financial problems or merged with larger

players. Some of the notable issuers to exit or during this period include ContiMortgage,

First Plus, Equicredit, The Money Store (acquired by First Union), and Green Tree

(acquired by Conseco). Other lenders were acquired by larger players, such as Advanta

(by JPMorgan), and Associates (by CitiMortgage). The financial problems among lenders

were mainly caused by a combination of lax underwriting standards, aggressive gain-on-

sale income accounting, and unfavorable market conditions after the liquidity crisis in 1998.

2002~2004 Expansion

The HEQ market expanded drastically since 2002. Some of the key drivers are:

1. Increased loan origination due to record high purchases and refinancing

motivated by historically low mortgage rates, and more cash-out financing

resulted from strong housing price appreciation.

2. Increasing use of securitization for funding and the advent of net interest margin

(NIM) technology. In 2001, issuers began to more regularly monetize the senior

component of residual cash flow in the form of a NIM security. This enables ABS

issuers to maximize deal issuance proceeds and reduce or eliminate residual risk.

The rapid growth of NIM securitizations prompted some dealer conduits to enter

the securitized HEQ market after 2001.

3. The subprime market has expanded to include borrowers that were traditionally

covered by Alt-A lenders.

31 March 2006


Using the Right Rating Performance Measures of SF Securities for CDO Analysis

87

The default rate is one of the most important parameters in rating CDO tranches. The loss

distribution of the underlying collateral is derived by incorporating a pre-determined default

matrix based on the historical default experience of corporate bonds, such as Moody’s

Idealized Loss Rates, and a loss model: Moody’s BET (Binomial Expansion Technique) or

CBM (Correlated Binomial) applies a single statistic (WARF) to generate the loss

distribution, while S&P’s Evaluator and Fitch’s VECTOR takes a simulation-based

approach, deriving the default probability of each asset from its rating.

Rating agencies periodically publish their default and loss studies or rating transitions of

structured finance (SF) securities. These results could serve as benchmarks to assess the

reasonableness of assumptions used by underwriters in modeling and structuring a new

SF CDO. Investors may also use these results to make asset allocation decisions, under

the assumption that these measures can separate the “good performers” from the “bad

performers” and that the future will follow the past.

Ideally, we should be able to find the desired figures from the agencies’ reports with ease.

In practice however, we often find ourselves swamped by all the tables and numbers and

the various methodologies used.

In this section, we discuss some of the nuances of rating agencies’ rating performance

measures and recommend the appropriate numbers to use for CDO analysis.

Things to keep in mind Based on our experience, we suggest checking at a minimum, the following points to correctly

understand rating transition numbers to be able to conduct true “apples-to-apples” analysis:

1. What products, such as HEL, resi-A, credit receivables, and etc., are included in each category, i.e. ABS, CMBS or RMBS?

2. Are the numbers calculated by cohort rating or original rating?

3. Are the numbers based on dollar amount or number of bonds?

4. What’s the frequency of the data; monthly, yearly, or other?

5. Are the numbers global or only US?

6. Are the numbers weighted or un-weighted, and if weighted, how?

7. Are the numbers adjusted for withdrawn ratings?

Even if these questions are answered, there still remains the most important question: how are the numbers actually calculated? We address this later.

87 This section was originally published in "The CDO Strategist", Issue #9, October 19, 2005.

31 March 2006


For the first question, all three agencies group SF securities into 3 categories: ABS, RMBS

and CMBS.88 However, Moody’s and Fitch put HEL/subprime mortgage into “ABS”, while

S&P puts HEL/subprime mortgage into “RMBS”.

As for the inclusion-universe of the rating transition numbers, we think numbers based on

original ratings are more suitable than those based on cohort ratings.89 Cohort rates tend

to over-estimate the actual numbers as a cohort includes seasoned bonds and defaults

tend to rise as loans season for most SF securities.90 As CDOs invest mostly in recently-

issued assets, cohort rates are not suitable.

Exhibit 78: Five-Year Cumulative Impairment Rates of Home Equity Bonds by Original and Cohort Rating, 1993-2004*

2.35%

6.99%

26.88%

41.13%

3.11%

14.09%

40.02%

51.82%

0%

10%

20%

30%

40%

50%

60%

A Baa Ba B

Original or Cohort Rate

5-Y

ea

r C

um

ula

tiv

e Im

pa

irm

en

t R

ate

Original Rating

Cohort Rating

* Moody’s Special Comment: “Default & Loss Rates of Structured Finance Securities: 1993-2004”, July 2005. Rates of Aaa and Aa rated are not

shown as they are all zero.

Source: Moody’s

As indicated by Exhibit 78, based on Moody’s 5-year cumulative impairment rates, the

difference between original rating-based rates and cohort rating-based rates may be very

significant: for Baa-rated bonds, the latter could double the former.

Moody’s Material Impairment Rate Since 2002, Moody’s conducts an annual study of the impairment rate of SF securities. As

Moody’s does not have a “default” rating, this study has significant implications in the

sense of “filling the gap”.91 An impairment rate includes uncured payment default, interest

shortfall or principal write-down, and downgrade to Ca or below. Although generally

speaking, the impairment rate is broader than default rate, we think it is the most

appropriate figure for CDO analysis for the following reasons:

1. Generally consistent with how defaults are defined in SF CDOs.

2. The only study, among all provided by rating agencies, that provides numbers based on original ratings in addition to cohort ratings.

88 Recently, they also added a CDO category.

89 A cohort includes all outstanding bonds issued up to (and including) the beginning of the cohort unit, such

as 1-year or 5-year cohort. 90

For a detailed discussion on cohort rates, please refer to: “The CDO Strategist - Default Assumptions for BBB HEQ in SF CDOs”, May 11, 2005. 91

The lowest rating of Moody's is "C", while S&P has a "D" rating which corresponds to "default".

Moody’s and Fitch

classify HEL into

ABS, while S&P

puts HEL into RMBS

For CDO analysis,

statistics based on

original ratings are

more meaningful

than those based on

cohort ratings

An impairment rate

includes uncured

payment default,

and downgrade to

Ca or below

31 March 2006


3. The methodology, by which the numbers are calculated, makes the most sense, in our view. A more detailed explanation and numerical examples are provided in Appendix I.

4. Comprehensive numbers, broken down by sectors (including HEL), ratings and

terms, are publicly available.

However, there are still some issues and nuances one needs to be aware of when

applying the impairment rates to CDO analysis.

How are default assumptions applied in CDO modeling? Let’s take a moment to briefly review how default assumptions are applied in the CDO

modelling process. Most equity marketing books or CDO term sheets have a chart similar

to Exhibit 80, which shows projected equity returns of a mezzanine (with collateral average

rating of BBB) SF CDO under different default rates. How should this chart be interpreted?

Exhibit 79: Illustrative Equity Returns by Default Rates (of a MZ SF CDO)*

15.1%

13.6%

12.0%

10.1%

8.1%

5.7%

2.9%

-0.2%

-3.5%

-7.1%

-10.8%

-15%

-10%

-5%

0%

5%

10%

15%

20%

0.00% 0.25% 0.50% 0.75% 1.00% 1.25% 1.50% 1.75% 2.00% 2.25% 2.50%

Annual Default Rate

Eq

uit

y R

etu

rn

* Calculated to 9-year auction call date. Recovery assumption is 60% with 1-year lag.


The “Annual Default Rate” is a constant annual default rate (CADR). Take the 0.5% CADR

as an example: this means that if the default rate of the underlying bonds stays the same

at 0.5% per year, the return to the equity holder, if calculated to the first auction call date 9

years later, will be 12%. One thing to keep in mind: the default rate is applied at the “bond”

level and to the remaining performing dollar balance.92

The 0.5% CADR implies a 5-year cumulative default rate of about 2.48%.93 Alternatively,

we can also specify a cumulative default rate first, say, 2.5% for 5 years. Then we need to

decide how to allocate this 2.5% over the 5-year period, i.e., the default timing. Rating

agencies will normally stress different default timing patterns – front-loaded, back-loaded

or evenly-distributed – during the rating process. We will discuss empirical evidences as to

the default timing later.

By the same token, Exhibit 80 shows a similar chart but for a high grade SF CDO. In this

hypothetical deal, a 0.1% CADR implies a return of 12.7%.

92 Although we always prefer to model the deal at the "loan level", meaning each loan underlying the HEL or

RMBS deal, in reality most of the new-issue SF CDOs are modeled at the bond level. 93

Based on the method discussed in Appendix I.

31 March 2006


Exhibit 80: Illustrative Equity Returns Varying by Default Rates (of a HG SF CDO)

14.3%

12.7%

11.0%

9.1%

6.9%

4.4%

1.7%

-1.2%

-4.4%

-7.8%

-11.5%

-15%

-10%

-5%

0%

5%

10%

15%

20%

0.0% 0.1% 0.2% 0.3% 0.4% 0.5% 0.6% 0.7% 0.8% 0.9% 1.0%

Annual Default Rate

Eq

uit

y R

etu

rn


We discuss some important points about the cumulative default rate:

1. Using different units of time to calculate the default rate, such as monthly, yearly,

or others, makes a huge difference: a 5-year default rate using 5-years as the unit

could be much lower than a 5-year default rate using an iterative process of

annual default rates, as discussed in Appendix I. This is similar to the

compounding effect in interest rate calculations.

2. A cumulative default rate could imply the following:

1) It could mean the percentage of assets defaulted during the time

period. For example, if there are 100 assets initially, a 5-year default rate

of 5% could mean that 5 assets defaulted over the 5-year period.

2) OR, it could also mean the probability for an asset to default over the

period. A 5-year default rate of 5% means, for each asset, there is a 5%

chance it will default by the end of the 5th year. Alternatively, it could

“survive” to the end of the 5th year with a probability of 95%.

It is crucial to understand these differences. As we will show below, different

rating agencies use different methods to calculate the numbers and the

conclusions could be dramatically different. For example, Moody’s calculation is

consistent with the second interpretation while S&P’s calculation is more in the

spirit of the first definition.

Using the impairment rates of the right sectors Due to adverse economic conditions, accounting and underwriting issues, and credit

deterioration in certain industries, some ABS sectors have suffered significant downgrades

and losses.

31 March 2006


Exhibit 81 shows the distribution of all securities impaired from 1993 to 2004 by ABS

sector. As we can see, the 3 worst-performing sectors in terms of impairments are MH,

Franchise Loans and Healthcare Receivables, of which almost 40% of the securities are

impaired. Furthermore, more than half of all impairments come from MH sector (263 out of

504 securities).

Given the significant exposure to these troubled sectors, many old vintage (1999-2002) SF

CDOs have also suffered disappointing performances. However, recent vintage SF CDOs

have stayed away from these sectors to the extent that most deals nowadays have zero

exposure to them. Instead, the majority of the collateral is invested in residential mortgage

related assets such as home equity and Resi-A. Thus, we think it is more relevant to look

at the impairment rates of residential mortgage sectors.

Exhibit 82 shows Moody’s 5-year cumulative impairment rates for US HEL, RMBS and

CMBS sectors based on the data from 1993 to 2004.94 There are several interesting

observations from these numbers:

1. For Aaa- and Aa-rated HEL and CMBS, there are NO impairments during any 5-

year period.

2. Moody’s rates are calculated based on the number of bonds, rather than the

dollar amount.

3. If assuming a high grade deal, with average underlying asset ratings of Aa and an

allocation of 50% HEL, 35% RMBS and 15% CMBS, the combined 5-year

cumulative impairment rate is about 0.51%.

4. Using the same allocation, but for a mezzanine SF CDO with average underlying

rating at Baa, the combined 5-year cumulative impairment rate will be about 6.7%.

5. For Ba-rated HEL, the impairment rate jumped significantly: from 6.99% at Baa

rating to 26.9% at Ba rating.

We believe the way Moody’s treats withdrawn ratings when calculating the numbers is

conservative (see Appendix I). The numbers without adjusting for withdrawals will be lower.

94 In its Special Comment: “Default & Loss Rates of Structured Finance Securities: 1993-2004”, July 2005,

Moody's for the first time published impairment rates of HEL alone.

Exhibit 81: Distribution of US ABS Impairments by Asset Type (1993-2004)

Asset Type

Number of Impaired

Securities from 1993-2004

Total Number of

Securities Studied* Percentage**

Manufactured Housing 263 662 39.73%

Franchise Loans 57 148 38.51%

Healthcare Receivables 12 32 37.50%

Aircraft & Equipment Leases 51 341 14.96%

Home Equity 95 3980 2.39%

Auto and Trucks 12 837 1.43%

Credit Card 13 1500 0.87%

Other ABS 1

Total 504 7500 6.72%

* Securities issued in 2004 are not included, and there are no impairments in 2004 vintage securities.

** These percentages could be viewed as the Lifetime Impairment Rates.

Source: Moody’s Special Comment: “Default & Loss Rates of Structured Finance Securities: 1993-2004”, July 2005, and “Default & Loss

Rates of Structured Finance Securities: 1993-2003”, September, 2004

31 March 2006


Exhibit 82: Moody’s 5-Year Cumulative Impairment Rates by Original Rating and Sector (1993-2004)

0.00% 0.00%

2.35%

6.99%

26.88%

1.02% 1.45% 1.20%

8.45%

6.05%

0.00% 0.00%0.66%

1.62%

3.75%

0%

5%

10%

15%

20%

25%

30%

Aaa Aa A Baa Ba

US HEL US RMBS US CMBS

Source: Moody’s Special Comment: “Default & Loss Rates of Structured Finance Securities: 1993-2004”, July 2005

Another reason the impairment rates at the Baa and Ba levels might seem higher than

expected is that Moody’s 2004 study changed the unit of measurement from calendar

years to months.95 As a result, the impairment rates are higher than those using calendar

years, according to Moody’s.

Finally, it is also interesting to take a deeper look at the impaired securities. In Exhibit 83

we list the number of impaired HEL bonds by vintage and original ratings. There are a

couple takeaways:

1. There have been no impairments for HEL bonds rated A1 or higher.

2. There have been no impairments for HEL bonds issued after 2001.

Default timing is just as important as the default rate. Moody’s publishes impairment rates

of 1-year up to 5-years. Based on their calculation (see Appendix I), we “back-out” the

“marginal impairment rates”, i.e., the impairment rate in each year given the bond has not

been impaired in the previous year.

95 And so going forward, according to Moody's.

Exhibit 83: Number of Impaired HELs by Vintage and Original Ratings (1993-2004)

Vintage A2 A3 Baa1 Baa2 Baa3 Ba1 Ba2 B2 B3 Total

1994 2 2

1995 1 2 1 4

1996 2 1 1 2 1 7

1997 1 2 3 6 3 4 19

1998 1 1 3 4 1 6 9 25

1999 3 3 2 4 12

2000 1 1 2 1 5

2001 1 2 3 6 12

Total 4 6 3 10 15 7 22 18 1 86


31 March 2006


Exhibit 84: Marginal Impairment Rates by Years Since Origination

0.00%0.15%

0.46%

2.66%

3.87%

0.08%

0.65%

1.56%

4.53%

1.94%

0.0%

0.5%

1.0%

1.5%

2.0%

2.5%

3.0%

3.5%

4.0%

4.5%

5.0%

1-Year 2-Year 3-Year 4-Year 5-Year

Years since origination

Ma

rgin

al

Imp

air

me

nt

Ra

te

Baa HEL Baa RMBS


Exhibit 84 shows our calculated marginal default rates of Baa-rated HEL and RMBS based

on Moody’s cumulative rates. Surprisingly, contrary to traditional wisdom, which holds that

default timing for HEL and RMBS securities is front-loaded, the Baa-rated HEL bonds

show a back-loaded default timing pattern – default rates appear to increase faster in later

years than in earlier years. Baa-rate RMBS exhibit a similar pattern except that its

marginal rate peaks in the 4th year and drops significantly in the 5th year.96

Although SF CDOs invest mostly in new-issue securities, they also invest in seasoned

bonds. So, for example, using these marginal rates, a one-year seasoned Baa-rate HEL

bonds should be assigned a default rate of 0.15% rather than 0% for the first year, 0.46%

instead of 0.15% for the second year and so on.

Other rating performance measures While all rating agencies, Moody’s, S&P and Fitch, publish performance measures such as

rating transitions, we think they are less applicable to SF CDO modeling. The main

reasons include:

1. All the rating transitions available are based on cohort ratings. As we discussed

previously, this limits their suitability for modeling CDOs.

2. Some rating transitions, such as S&P’s, are calculated based on “rolling cohorts”

and the term of the cohort depends on the term of the transition rate. For example,

to calculate the 5-year transition, a 5-year cohort needs to be formed. Thus, a

bond must be 5 years seasoned for it to be included in the calculation of the 5-

year transition rate. While this approach is easier to understand, it misses data for

the most recent 4 years when calculating a 5-year transition rate.97

96 Unfortunately, Moody's only provide cumulative rates up to 5 years. Otherwise we would be able to

derive marginal rates for more seasoned bonds. 97

A bond issued, say, 4 years ago, can not be used to calculate a 5-year transition rate since it does not have performance history longer than 5 years.

31 March 2006


We deem rating transition studies very important for ad hoc performance reviews and

comparisons. One common argument is SF securities are more stable rating performers

than corporate bonds.

As shown in Exhibit 85 and Exhibit 86, based on both Moody’s and S&P’s annual rating

transitions, HEL/RMBS outperforms corporate bonds as they have a much lower chance

of being downgraded across most rating categories. Fitch’s results also shares similar

conclusions.

Final Thoughts Moody’s ratings are ultimately determined by the expected loss rate, rather than the

default rate. Moody’s Idealized Loss Rates, which are based on the default experience of

corporate bonds, are used to derive the default rates of SF securities for SF CDOs. These

corporate loss rates are also calculated based on cohort ratings.

We think a better measurement and benchmark system of default rates needs to be

developed for SF CDOs. The rating performance studies conducted by rating agencies are

certainly valuable for achieving this goal, but it remains to be seen how the current study

results can be incorporated into the rating processes. In our view, to date, Moody’s

Impairment Rates are the most suitable.

The new year is coming, and here is our resolution: an ideal measurement of SF securities

default rate, calculated based on dollar amount and original ratings, separated by different

sectors, and with the default definition as close as possible to CDO standards. Both

marginal and cumulative rates are needed. To achieve this, further studies by rating

agencies are needed.

Exhibit 85: Downgrade Rates from Moody’s Annual Ratings Transition Matrices*

0.1%0.5%

1.7%

3.7%

8.1%8.4%

0.4%

2.0% 2.1%

3.0%

3.8%

4.9%

8.3% 8.5%

6.4% 6.4%

10.5%10.9%

0%

2%

4%

6%

8%

10%

12%

Aaa Aa A Baa Ba B

HEL Downgrade Rate RMBS Downgrade Rate Corporate Downgrade Rate

* By COHORT. For corporate bonds, the sample period is 1983-2004; for HEL and RMBS, the sample period is 1990-2004

Source: Moody’s Special Comment:: “Structured Finance Rating Transitions: 1983-2004”, February 2005

31 March 2006


Exhibit 86: Downgrade Rates from S&P’s Annual Ratings Transition Matrices*

0.2%

1.5%1.2%

1.8%

2.8%

4.5%

8.3%8.9%

6.5%6.0%

10.3%

11.0%

0%

2%

4%

6%

8%

10%

12%

AAA AA A BBB BB B

RMBS Downgrade Rate Corporate Downgrade Rate

* By COHORT. S&P’s RMBS include subprime mortgage transactions

Source: S&P: “Global Structured Securities Rating Performance: 1978-2004”, March 2005

Appendix I. Moody’s Impairment Rates by Original Ratings To explain how Moody’s calculates its cumulative impairment rates by original ratings, we

use an example. Here, we show how a 3-year rate is calculated.98

Exhibit 87: Illustration of Calculating Moody’s Cumulative Impairment Rates by Original Rating

Year 1 Year 2 Year 3

End End End Rating Beginning

Impaired Withdrawn

Rating Beginning

Impaired Withdrawn

Rating Beginning

Impaired Withdrawn

Vintage Year 1

Baa 100 0 10 Baa 88 2 10 Baa 74 4 10

Ba 2 1 0 Ba 2 1 0

B 1 1 0

Vintage Year 2

Baa 100 4 8 Baa 86 0 10

Ba 2 1 0

Vintage Year 3

Baa 200 2 20


Suppose the sample period starts from Year 1 and there were 100 Baa-rated bonds at the

beginning of Year 1, which we call “Vintage Year 1”. The top block in Exhibit 87 tracks the

status of these 100 bonds during the 3-year period. For example, at the beginning of Year

3, only 74 bonds remain at Baa, 4 of which became impaired and 10 of which were

withdrawn in Year 3.

98 This example is purely hypothetical and for illustration only.

31 March 2006


At the beginning of Year 2, 100 more Baa-rated bonds were issued which we call “Vintage

Year 2”, and the middle block in Exhibit 87 tracks the status of these 100 bonds in the next

2 years.

At the beginning of Year 3, 200 more Baa-rated bonds were issued which we call “Vintage

Year 3”, of which 2 were impaired and 20 were withdrawn in one year.

To calculate the first year Baa-impairment rate, all the numbers highlighted in yellow are

included, i.e., the first-year experience of all the bonds in 3 vintages are used. The

withdrawn ratings are adjusted by taking half of the withdrawals off the denominator. The

weighted average first-year marginal impairment rate is calculated as:

(0+4+2)/(100+100+200-10/2-8/2-20/2) = 1.57%

To calculate the second year marginal Baa-impairment rate, all the numbers highlighted in

pink are included. Therefore, numbers from only 2 vintages, Vintage Year 1 and Vintage

Year 2, are used. It is calculated as follows:

(2+0)/(88+86-10/2-10/2) = 1.22%

By the same token, the third year marginal Baa-impairment rate can be calculated by

using only the numbers of Vintage Year 1, highlighted in gray. It is calculated as:

4/(74-10/2) = 5.8%

The cumulative rates are calculated using an iterative process. The 2-year cumulative rate

is calculated as:

1-(1-1.57%)*(1-1.22%) = 2.78%

A 2-year survival rate is calculated first, (1-1.57%)*(1-1.22%), and then the 2-year

cumulative impairment rate is 1 minus the survival rate.

Similarly, the 3-year cumulative rate is calculated as:

1-(1-1.57%)*(1-1.22%)*(1-5.8%) = 8.41%

This methodology is also used for calculating the impairment rates by cohort ratings and is

consistent with how Moody’s derives its corporate default rates.99

We view this as a very reasonable approach. However, there is one downside when using

original ratings. For example, notice that there are 2 bonds in Vintage Year 1 that were

downgraded to Ba in Year 1 (in Exhibit 87, there are 2 bonds which begin Year 2 at the Ba

rating), and these 2 bonds will not be included in the calculation of the impairment rates of

Ba rating as they are NOT originally rated Ba. Thus, some useful information might be lost

among the calculations.

99 Please see Moody's Special Comment: "Default & Recovery Rates of Corporate Bond Issuers: 1970-

2001", February 2002.

31 March 2006


Appendix II. List of Select Publications on Rating Performance by Rating Agencies

Exhibit 88: Select Publications on Rating Performance by Rating Agencies

Rating Agency Report Title

Publication

Date

Moody's Default & Loss Rates of Structured Finance Securities: 1993-2004 July-05

Moody's Default & Loss Rates of Structured Finance Securities: 1993-2003 Sep-04

Moody's Payment Defaults And Material Impairments of U.S. Structured Finance Securities: 1993-2002 Dec-03

Moody's Structured Finance Rating Transitions: 1983-2004 Feb-05

Moody's Structured Finance Rating Transitions: 1983-2003 Feb-04

Moody's Structured Finance Rating Transitions: 1983-2002 Jan-03

Standard & Poor's Global Structured Securities Rating Performance: 1978-2004 Mar-05

Standard & Poor's Structured Finance Global Ratings Roundup Quarterly Quarterly

Fitch Fitch Ratings 1991-2004 Structured Finance Transition Study Mar-05

Fitch Fitch Ratings 1991-2003 Structured Finance Transition Study Nov-04

Fitch Structured Finance Rating Transition Study May-02

Source: Credit Suisse, Moody’s, S&P, Fitch

31 March 2006


Impact of S&P’s New Rating Criteria on SF CDOs

100

Standard and Poor’s recently revised its proprietary CDO modeling tool, CDO Evaluator, from Version 2.4.3 (E2) to Version 3.0 (E3). For now (January 2006), the new Evaluator is only being applied to synthetic CDOs with no excess spread. However, as S&P has indicated, the same model will be applied to cash CDOs, hybrid CDOs, and synthetic CDOs with excess spread too, most likely in early 2006 when additional cash flow criteria are finalized. It is crucial for the CDO market to understand the implications of the new methodology. In this section, we focus on the impact of E3 on SF CDOs.

New default rate assumptions for ABS securities One of the key changes in E3 with potential implications on SF CDOs is the new default

rate assumption for ABS securities. There is no change in ABS securities’ correlation

assumptions and the recovery rate changes have minimal impact as the ABS recovery

rates are user defined in the model.101

Previously in E2, the ABS default rates were derived by using corporate default rates as a

proxy and the rates were one-dimensional – i.e., by ratings only, regardless of maturity. E3

uses a two-dimensional default matrix – by rating and by maturity. According to S&P,

this matrix is built based on historical transition rates. By comparing the numbers, we

made the following observations:

1. S&P assumes an upward sloping default curve by maturity. The new model

gives benefit to shorter maturity securities by assuming a relatively steep default

curve, i.e., the default rates for shorter maturities are much lower than the rates

for longer maturities. Exhibit 89 shows BBB default rates as an example - the

new default rate starts very low at a 1-year maturity and converges to the old (E2)

default rate at 2%.

2. Compared to the previous default curve, depending on maturity, the new

default rates could be higher or lower. At longer maturities, for ratings

above BBB, the new default rates are much lower; while below BBB, the

new default rates are much higher; for shorter maturities, all new default

rates are lower. Exhibit 90 shows the 7-year cumulative default curve by rating

in E3 as much steeper than E2.102 However, the 5-year cumulative default curve

falls below the previous curve in E2 almost across all ratings. S&P capped the

default rate of ABS securities at the 7th year, i.e., after year 7, the cumulative

default rate does not increase and thus there are no additional defaults (i.e., the

marginal default rate is assumed to be zero after 7 years). The “bad” news is,

because legal maturities are used in E3 and ABS securities have typically very

long legal maturities, the 7-year default rates are almost always used. So in reality,

the “credit” given to shorter securities may never be “cashed”. All these nitty-gritty

details will have profound impact on SF CDO ratings – both in mezzanine and

high grade deals – we discuss more in detail later.

100 This section was originally published in "The CDO Strategist", Issue #13, January 25, 2006.

101 In E3, stochastic recoveries can be used in addition to constant recoveries for corporate bonds or loans.

But for ABS securities, there is no change regarding recovery. 102

A detailed default matrix can be found in Appendix I.

31 March 2006


Exhibit 89: BBB Default Rates by Maturity in Evaluator 3 & Evaluator 2

0.0%

0.5%

1.0%

1.5%

2.0%

2.5%

1 2 3 4 5 6 7

Maturity

Cu

mu

lati

ve

De

fau

lt R

ate

BBB Cum Default Rate in E3

BBB Default Rate in E2


Exhibit 90: Default Rates of ABS in Evaluator 3 and Evaluator 2

0%

2%

4%

6%

8%

10%

12%

14%

16%

AAA AA+ AA AA- A+ A A- BBB+ BBB BBB- BB+ BB BB-

Rating

Cu

mu

lati

ve

Defa

ult

Rate

E2 ABS Default Rate

E3 ABS Default Rate (7 year)

E3 ABS Default Rate (5 year)


31 March 2006


New default rate assumptions of CDO tranches The CDO default rate matrix in CDO Evaluator serves two purposes:

1. It determines the default rate to be used for each CDO tranche in the underlying

asset pool when assessing their default risk.

2. It also determines the subordination levels of the CDO being rated – it can serve

as a confidence level to find the cutoff point – the Scenario Default Rate (SDR) –

and consequently the Scenario Loss Rate (SLR) and the subordination levels of

each tranche. For example, if we are to rate a tranche “AAA” and the associated

default rate of an “AAA” rating is 0.5%, the SDR is determined such that the

chance of the default rate of the underlying collateral being higher than this SDR

is less than or equal to 0.5%. All things equal, the higher the default rate of the

target rating (i.e., more defaults are allowed to achieve the target rating), the

lower the cutoff point (the SDR) and thus the lower the subordination required.

Previously the CDO default rate assumptions were the same as that of corporates. In E3,

for CDO ratings above BBB, the new default rates are lower, while for BBB and below

ratings, the new default rates are higher. Exhibit 91 shows 8-year default rates as an

example.103

Exhibit 91: 8-Year CDO Default Rate in Evaluator 3 and Evaluator 2

0%

5%

10%

15%

20%

25%

30%

35%

40%

45%

50%

AAA AA A+ A- BBB BB+ BB- B

Rating

Defa

ult

Rate

8-year CDO Default Rate in E3

8-year CDO Default Rate in E2


There are two forces working against each other here as the same CDO default matrix is

used for two purposes: on one hand, a higher (lower) default rate used for the underlying

CDO tranches results in higher (lower) default risk of the underlying CDO tranches in the

collateral pool and thus raises the overall default risk of the collateral pool (and ultimately it

results in higher (lower) subordination requirement); on the other hand, higher (lower)

default rates used for the target CDO being rated also suggests a more generous

standard – more defaults are allowed to achieve the target rating – and results in lower

(higher) SDR and subordination. It seems that the first force outweighs the second, as

suggested by the example we show next.

103 A detailed default matrix can be found in Appendix I.

31 March 2006


First example – a mezzanine SF CDO First we ran both E3 and E2 on a real mezzanine SF CDO issued in 2005. Exhibit 92 and

Exhibit 93 show the rating and sector breakdown of this deal. The underlying portfolio of

this deal is typical for recent SF CDOs. The weighted average rating is about BBB+/BBB

and the vast majority of the collateral is invested in Resi B&C and Resi A.

Exhibit 92: Rating Breakdown (Sample MZ SF CDO)

Exhibit 93: Sector Breakdown (Sample MZ SF CDO)

A-

7%

AA-

2%

AAA

1%

BBB

22%

BBB-

21%

BBB+

29%

A

10%

A+

2%

BB+

3%

BB

1%

AA

2%

RMBS B&C

64%

CDOs

3%

Aerospace &

Defense

1% CMBS (Large Loan,

Single Borrower, and

Single Property)

1%

ABS Consumer

4%ABS Commercial

1%

RMBS A

13%

CMBS Diversified

(Conduit and CTL)

5%

Manufactured

Housing

2%

REITs and REOCs

6%

Source: Credit Suisse. S&P Source: Credit Suisse, S&P

We use these assets as inputs to run this deal through both E3 and E2. We compare

SDR’s from both models in Exhibit 94 and, with the exception of AAA and AA+ ratings, E3

generates higher SDR’s for all other rating categories. The main reason for this is because

of the presence of BBB- and below rated assets in the pool – about 25% in total – which is

penalized by E3 as the default rates of ratings below BBB are higher in the new model.

While the default rates of BBB+ and above ratings are lower and the pool does have more

than 50% BBB+ and above rated assets, it seems that it is not enough to overcome the

penalty given to the lower rated assets.104

“Rating Default Probability” (RDP) is determined by rating and maturity from the CDO

default matrix. As we run the analysis to the auction call date of 8 years, the RDP’s are

exactly the same as the numbers in Exhibit 91. Exhibit 96 shows the default rate

distributions from both E3 and E2. It turns out that the default distribution from E3 has a

fatter right tail. Translation: due to the higher default rates assumed for lower-rated assets

in E3, the model estimates that the underlying portfolio will have a higher chance of

experiencing high default rates.

104 In E2, S&P uses a set of stress factors for SDR. For example, a stress factor of 1.2 is applied to AAA

rating to reach the final SDR. In E3, the stress factors have been removed.

31 March 2006


Exhibit 94: Comparison of SDR – Mezz. SF CDO

Desired

Rating

E3 Rating Default

Probability SDR_E3

E2 Rating Default

Probability SDR_E2

SDR_E3/SDR_E2

Ratio

AAA 0.405% 18.04% 0.658% 18.93% 0.95

AA+ 0.584% 16.54% 0.835% 16.77% 0.99

AA 0.927% 14.64% 1.445% 13.75% 1.06

AA- 1.182% 13.69% 1.650% 12.84% 1.07

A+ 1.472% 12.84% 1.896% 11.91% 1.08

A 1.774% 12.12% 2.204% 10.99% 1.10

A- 2.395% 10.96% 2.632% 10.01% 1.09

BBB+ 3.413% 9.70% 3.492% 8.74% 1.11

BBB 5.310% 8.20% 4.667% 7.52% 1.09

BBB- 9.891% 6.23% 7.360% 5.92% 1.05

BB+ 12.007% 5.67% 11.525% 4.39% 1.29

BB 16.810% 4.72% 15.419% 3.57% 1.32

BB- 22.544% 3.95% 17.816% 3.14% 1.26


Exhibit 95: Comparison of SLR – Mezz. SF CDO

Desired

Rating

E3 Rating Default

Probability SLR_E3

E2 Rating Default

Probability SLR_E2

SLR_E3/SLR_E2

Ratio

SLR_E3 –

SLR_E2

AAA 0.405% 13.33% 0.658% 13.85% 0.96 -0.52%

AA+ 0.584% 11.56% 0.835% 11.59% 1.00 -0.03%

AA 0.927% 10.27% 1.445% 9.55% 1.08 0.72%

AA- 1.182% 9.63% 1.650% 8.93% 1.08 0.70%

A+ 1.472% 8.47% 1.896% 7.73% 1.10 0.74%

A 1.774% 8.01% 2.204% 7.15% 1.12 0.86%

A- 2.395% 7.26% 2.632% 6.52% 1.11 0.74%

BBB+ 3.413% 5.86% 3.492% 5.17% 1.13 0.69%

BBB 5.310% 4.97% 4.667% 4.45% 1.12 0.52%

BBB- 9.891% 3.79% 7.360% 3.52% 1.08 0.27%

BB+ 12.007% 3.14% 11.525% 2.36% 1.33 0.78%

BB 16.810% 2.62% 15.419% 1.92% 1.36 0.70%

BB- 22.544% 2.19% 17.816% 1.69% 1.29 0.50%


Exhibit 96: Default Rate Distribution in E3 and E2

0 %

5 %

1 0 %

1 5 %

2 0 %

2 5 %

3 0 %

3 5 %

4 0 %

4 5 %

0 % 2 % 4 % 6 % 8 % 1 0 % 1 2 % 1 4 % 1 6 % 1 8 % 2 0 % 2 2 %

D e fa ult R a te

Probablity

P ro b a b i lity o f D e fa ult Ra te _ E 3

P ro b a b i lity o f D e fa ult Ra te _ E 2


31 March 2006


Based on S&P’s methodology, cash flow analysis is required to verify that each rated

tranche can withstand defaults up to its SDR.105 Ultimately, the credit enhancement or

subordination levels are determined by a Scenario Loss Rate (SLR). The SLR determines

the subordination level such that the probability of loss of the underlying portfolio

exceeding the subordination – i.e., the SLR – is no greater than the default probability of

the target rating. This is consistent with S&P’s basic scheme of a rating – it reflects the

probability of “first dollar loss”.106 To determine SLR, recovery assumptions need to be

made.107 Exhibit 95 lists the SLR’s from both E3 and E2. Similar to the pattern of SDR’s in

Exhibit 94, the SLR’s – the credit enhancement levels – are higher for most of the ratings

except AAA and AA+ based on E3. For example, for BBB rating, the enhancement level

will have to be raised by 52 bps if E3 is used.

Second example – a high grade SF CDO Because the new default rates of ABS securities rated above BBB are lower than those in

E2, one would expect the subordination levels required by E3 for high grade SF CDOs to

be lower than those required by E2. A sample run through a real high grade SF CDO

issued in 2005 proves so.

Exhibit 97: Rating Breakdown Exhibit 98: Sector Breakdown

Sample HG SF CDO Sample HG SF CDO

A

7%

A-

5%

A+

2%

AA

27%

AA-

11%

AA+

9%

AAA

39%

ABS Commercial

3%

CDOs

31%

CMBS Diversified

(Conduit and CTL)

1%

Monoline/FER

Guaranteed

4%RMBS A

8%

RMBS B&C, HELs,

HELOCs, and Tax

Lien

53%

Source: Credit Suisse. S&P Source: Credit Suisse, S&P

As shown in Exhibit 97 and Exhibit 98, this deal has a weighted average rating of AA+/AA

with no collateral rated below A-. The portfolio is more than 60% invested in Resi B&C

and Resi A combined and over 30% in CDOs.

105 A Breakeven Default Rate – the maximum default percentage the transaction can withstand without any

loss to the rated tranche, determined based on cash flow analysis. For a tranche to be rated at the target rating, the SDR has to be lower than the Breakeven Rate. 106

Moody's ratings are determined based on expected loss. 107

Please refer to Appendix II for details.

31 March 2006


Exhibit 99: Comparison of SDR – High Grade SF CDO

Desired

Rating

E3 Rating Default

Probability SDR_E3

E2 Rating Default

Probability SDR_E2

SDR_E3/SDR_E2

Ratio

AAA 0.405% 5.00% 0.658% 7.90% 0.63

AA+ 0.584% 4.45% 0.835% 6.94% 0.64

AA 0.927% 3.79% 1.445% 5.58% 0.68

AA- 1.182% 3.49% 1.650% 5.16% 0.68

A+ 1.472% 3.21% 1.896% 4.76% 0.67

A 1.774% 2.98% 2.204% 4.36% 0.68

A- 2.395% 2.61% 2.632% 3.93% 0.66

BBB+ 3.413% 2.23% 3.492% 3.35% 0.66

BBB 5.310% 1.80% 4.667% 2.82% 0.64

BBB- 9.891% 1.16% 7.360% 2.16% 0.54

BB+ 12.007% 1.01% 11.525% 1.53% 0.66

BB 16.810% 0.81% 15.419% 1.18% 0.69

BB- 22.544% 0.46% 17.816% 0.99% 0.46


Exhibit 100: Comparison of SLR – High Grade SF CDO

Desired

Rating

E3 Rating Default

Probability SLR_E3

E2 Rating Default

Probability SLR_E2

SLR_E3/SLR_E2

Ratio

SLR_E3 –

SLR_E2

AAA 0.405% 2.51% 0.658% 4.03% 0.62 -1.52%

AA+ 0.584% 1.96% 0.835% 3.08% 0.63 -1.13%

AA 0.927% 1.67% 1.445% 2.47% 0.68 -0.80%

AA- 1.182% 1.53% 1.650% 2.29% 0.67 -0.76%

A+ 1.472% 1.12% 1.896% 1.68% 0.67 -0.56%

A 1.774% 1.04% 2.204% 1.54% 0.68 -0.50%

A- 2.395% 0.92% 2.632% 1.39% 0.66 -0.47%

BBB+ 3.413% 0.66% 3.492% 1.00% 0.66 -0.34%

BBB 5.310% 0.52% 4.667% 0.84% 0.62 -0.32%

BBB- 9.891% 0.36% 7.360% 0.64% 0.56 -0.28%

BB+ 12.007% 0.28% 11.525% 0.40% 0.70 -0.12%

BB 16.810% 0.21% 15.419% 0.31% 0.68 -0.10%

BB- 22.544% 0.13% 17.816% 0.26% 0.50 -0.13%


Both the SDR and SLR are significantly lower across all ratings if E3 is used, as shown in

Exhibit 99 and Exhibit 100. Take BBB SLR as an example: E3 requires only 2/3 of the SLR

required in E2. Alternatively, in terms of subordination level required, it is 32 bps lower.

For high grade deals with very high leverage, this kind of drop would be very significant.

31 March 2006


Impact on the secondary market Adopting a new version of Evaluator by S&P may have a significant impact on the

secondary SF CDO market, both synthetic and cash flow deals. All things equal, high

grade deals are more likely to be upgraded than mezzanine deals because of this change

in the model, and vice versa for downgrades. For existing synthetic deals without excess

spread, the agency has already put some deals on watch list and also indicated no

additional rating actions are expected at this time. However, in the future, because of

credit migration in the underlying portfolio, investors need to understand the change in E3

in order to better evaluate their CDO holdings. For example, if there is some credit

deterioration to below BBB ratings in the pool, it will be even more likely for the deal to be

downgraded than before (with E2). For cash flow deals and synthetic deals with excess

spread, even though the agency is still finalizing the cash flow criteria and the ultimate

impact of E3 remains uncertain, investors still need to be aware of the potential changes

and take them into account in their evaluations.

In terms of rating distribution, the new model encourages a more bullet-like portfolio

versus a barbell portfolio for mezzanine SF CDOs. In other words, a portfolio of, say, all

BBB-rated assets will be required to have lower subordination levels than a half BBB+/half

BBB- portfolio, given the reasons we discussed earlier. This could have spread

implications on the underlying collateral markets. Of course, this will be a self-correcting

process, i.e., if the spreads of lower-rated assets widen to certain levels, their

attractiveness may outweigh the required additional enhancement.

Final comments As mentioned previously, S&P is still finalizing the cash flow assumptions such as interest

rate, amortization speed, etc. The final rating and subordination levels will be determined

by both the outputs from the new CDO Evaluator and the cash flow analysis. For example,

even though we found that the subordination levels of high grade SF CDOs based on E3

are lower than those from E2, we also expect that the new set of cash flow assumptions to

be released later might mute the impact to some degree.

31 March 2006


Value Shifting to Mezzanine SF CDOs108

In the past two months or so, subordinate (Baa1-Baa3) home equity (HEL) bonds spreads

have significantly widened. The widening started with the CDS/synthetic spreads on HEL

bonds, and now cash spreads have followed. Baa2 cash spreads stand at 235 basis

points, 100 bps wider than the level just 2 months ago, and Baa3 cash spreads have

doubled to 350 bps (see Exhibit 101).

With the collateral of recent mezzanine SF CDOs dominated by Baa2/Baa3-rated HELs,

the spread widening has had significant impact on the CDO market. We have seen new-

issue mezzanine ABS CDO liability spreads widen out, especially at the BBB level. In a

recent pricing, the BBB class was priced at L+400 bps. Given the recent re-pricing, we

believe that value is starting to shift to mezzanine ABS CDOs. In this issue’s Insight, we

share some of our thoughts and demonstrate reasons for our view.

Widening asset spread boosts equity IRR of Mezz. ABS CDOs Given the current spreads of HEL bonds (cash or synthetic) and the predominance of HEL

collateral in recent ABS CDOs, it is likely to have collateral pools with a weighted average

spread of around L+240 bps.109 110 Even with the recent widening in mezz. ABS CDO

liability spreads, the potential IRR for the equity tranche is still very attractive. We use a

hypothetical but representative mezz. ABS CDO to show some numerical examples.

Exhibit 102 shows the capital structure of our sample deal. Note we increase the Baa2

tranche spread to 400 bps. By assuming a 0.5% constant annual default rate (CADR) and

a 60% recovery rate, we calculate the IRR on the equity tranche to be 24.5%, significantly

higher than 10%-15% baseline numbers mezzanine ABS CDOs issued during the better

part of 2005.

108 This section was originally published in "The CDO Strategist", Issue #12, December 15, 2005.

109 The weighted average rating factor (WARF) of HEL bonds in a Mezzanine ABS CDO is around 360-420,

or Baa2/Baa3-rated. And HEL bonds could take 60-80% of the collateral, with the rest invested in Resi-A (10%), junior tranches of other CDOs (10%), and others such as CMBS or credit card receivables. 110

Note that a WAS of L+240 is possible for new deals initiating their ramp-up during the last few weeks. For deals already significantly ramped prior to the spread widening, a lower WAS is expected.

Exhibit 101: HEL CDS Spreads vs. Cash Spreads

December 9, 2005 2 Month Ago

Rating Cash Spread

Synthetic

Spread Basis Cash Spread

Synthetic

Spread Basis

Baa1 170 160 -10 120 115 -5

Baa2 235 230 -5 135 130 -5

Baa3 350 340 -10 175 200 25


Exhibit 102: Capital Structure of a Hypothetical Mezzanine ABS CDO*

Tranche Balance % Rating Coupon OC Target IC Target

A 400,000,000 80.0% Aaa L + 35

B 50,000,000 10.0% Aa2 L + 65 103.5% 115.0%

C 27,500,000 5.5% Baa2 L+ 400 101.0% 110.0%

Equity 22,500,000 4.5%

* We also assume this deal has a reinvestment/non-call period of 3 years, and an auction call date of 8 years after closing. We also

assume there is a turbo feature on the Baa2 tranche and the equity return is capped at 18% during the turbo period.


31 March 2006


But does this mean Mezz. ABS CDOs offer value? We think so Currently the baseline IRR expectation for CLO equity and high-grade SF CDOs falls into

the 10-15% range. However, just because the potential IRR of mezz. SF CDOs under one

particular set of assumptions is higher does not necessarily mean mezz. SF CDOs offer

superior value versus CLOs and/or HG SF CDOs. One may argue that the spread

widening in HEL is justified: the US housing market may slow down dramatically causing

defaults and losses among HEL bonds, which may increase and eventually ripple though

to mezz. SF CDOs. In other words, the 0.5% CADR we assumed is too low and the 24.5%

IRR will not be achieved.

There is certainly some merit to this argument. To check its validity, we approach this

issue from a slightly different angle: instead of comparing spreads and IRRs, we calculate

the implied level of risk the market is pricing in.

To talk relative value, we need to find a common benchmark. Before the recent re-pricing in

subordinate HEL bonds and mezzanine SF CDOs, the baseline equity IRRs of HY CLOs,

HG SF CDOs and mezz. SF CDOs were all hovering around 10-15%. Equity investors have

generally accepted these as IRR market target range. In a benign credit environment with

tight spreads across most markets, equity investors seem to agree on this IRR range as an

“equilibrium” level, where the risks they are taking are fairly compensated, and marginal

investors, who can switch among these different CDO products, are indifferent when

choosing which product to invest in terms of risk versus return trade-off. As the expected

IRRs of CLO and HG SF CDO equity has not changed significantly, it is not un-reasonable to

assume an ”equilibrium” IRR of 15%, i.e., at this IRR, equity investors will be indifferent to

choosing among CLOs, HG SF CDOs and mezz. SF CDOs.

Now let’s ask the following question: if 0.5% CADR is too low and an IRR around 25%

cannot be achieved, what is the right CADR to use in order to achieve a 15% IRR, for

mezz. SF CDOs?

We use the same hypothetical deal and solve for the CADR that would result in an IRR of

15%.

Exhibit 103: Implied CADR and Cumulative Default Rate

Recovery Rate

BBB CDO

Spread Equity IRR Imp. CADR

Imp. 5-y Cum

Default

Imp. 5-y Cum

Loss

60% 350 15% 2.67% 12.66% 5.06%

60% 400 15% 2.60% 12.34% 4.94%

60% 450 15% 2.53% 12.03% 4.81%

60% 400 10% 3.22% 15.10% 6.04%


As shown in Exhibit 103, the CADR has to reach 2.6% for the equity IRR to drop to 15%.

Let’s put this number into historical perspective: based on Moody’s Impairment Rate study,

the 5-year cumulative impairment rate of HEL is about 7%. However, if we calculate the 5-

year cumulative default rate of 2.6% CADR using the same methodology that Moody’s

uses, the cumulative default rate would be 12.34%, much higher than the historical 7%

cumulative impairment rate.111 Note that Moody’s 5-year cumulative impairment rate of

HEL is already a rather conservative number due to the following reasons:

• The Impairment Rate is, in general, a broader concept than default rate.

111 The 5-year cumulative number is calculated as one minus the 5-year survival probability, or 1-(1-

CADR)^5.

31 March 2006


• Moody’s 5-year cumulative rate is derived by calculating the marginal rates first.

However, when calculating the 4th or 5th year marginal rates, recent vintages (after

2001) of HEL are not included. The HEL sector has evolved considerably since

2001: it has gone through the pre-1996 nascent stage, the initial growth from

1996 to 1998, the consolidation period from 1999 to 2001, and dramatic

expansion since then. Many impairments due to industry-wide issues before 2001

are reflected in the 4th and 5th year marginal rates. As shown in Exhibit 104, there

is a huge jump from the 3rd year marginal rate (0.46%) to the 4th and 5th year

marginal rates (2.66% and 3.87%, respectively), and we believe this is due to the

sampling issue just discussed. More important, we believe the 5-year cumulative

impairment rate using these marginal rates over-estimates the actual experience.

Exhibit 104: Marginal Impairment Rates by Years Since Origination

0 .00%0.15%

0.46%

2.66%

3.87%

0.08%

0.65%

1.56%

4.53%

1.94%

0.0%

0.5%

1.0%

1.5%

2.0%

2.5%

3.0%

3.5%

4.0%

4.5%

5.0%

1-Year 2 -Year 3 -Year 4-Year 5 -Year

Y e ars s in ce o r ig in a tio n

Ma

rgin

al

Imp

air

me

nt

Ra

te

Baa HEL Baa RMBS


The fact that Moody’s 5-year Impairment Rate is over-estimated and our implied 5-year

cumulative default rate is even higher than the Moody’s number leads us to believe that

the market would over-estimate the default risk if assuming an equity IRR of only 15%. In

other words, we believe the actual default rate will be lower and the equity can achieve a

higher IRR than 15%. Thus, we find relative value in equity tranches of recent mezzanine

SF CDOs (priced in the past 2 weeks or so) and those still in the pipeline if market

conditions stay unchanged.

Some might contend that the equity IRR is high because the debt tranches are priced too

rich, i.e., the spread is too low. We increased the BBB tranche spread from 400 bps to 450

bps and re-ran the numbers. Both implied CADR and cumulative default rate dropped

slightly (see Exhibit 103) and the conclusion stands. We can certainly bump up the spread

further, but note that historically the highest BBB mezz. SF CDO tranche spread was 450

bps during early 2003 and the conclusion still won’t change much even if spreads are

raised. As a matter of fact, we find the BBB tranche attractive at L+400 to 450 level, even

more so if there is a turbo feature on the BBB tranche.

What if the spreads of senior tranches of SF CDOs jump as well? Given that the HEL

spreads have widened and stayed wide, we do believe the senior spreads will eventually

widen as well. We have already seen a deal price in the market with senior AAA at L+32

bps, 5-6 bps wider than recent historical lows. But notice that in the sample deal we used,

we already consider the spread widening of senior tranches by applying an aggregate

(senior and junior) AAA spread of 35 bps. Even if the spread widens further, say AAA

spread doubles to 70 bps, we can still achieve an equity IRR above 20% at 0.5% CADR. If

this does materialize, we would call for value in senior tranches given that most other AAA

tranches are priced at the high 20’s to low 30’s, all else being equal.

31 March 2006


There is something ELSE about mezz. SF CDOs In addition to the points just discussed, we like new mezz. SF CDOs also for the following

reasons:

• Because of the spread widening in subordinate HEL bonds, there is more leeway

now for SF CDO managers to find undervalued bonds and pick the right credits

for their deals.

• The more active synthetic market on subordinate HEL bonds and the attractive

synthetic spreads also help managers find value.

• The widening collateral spreads also allow mezz. SF CDOs to invest in just

floating bonds, instead of using fixed bonds to boost equity returns (as some HG

SF CDOs do) but also introduce other risks such as convexity risk.

Final Thoughts The fundamental reason we find value in new mezz. SF CDOs is because of the spread

widening in subordinate HEL bonds. The wider spreads allow the CDO to take advantage of

the arbitrage between asset and liability spreads and brings an attractive return, without

sacrificing credit quality or increasing leverage.

The widening new issue mezz. SF CDO spreads may also have an impact on secondary

bonds. If wider new issue spreads are used to price seasoned bonds which cause those

bonds to trade at a discount, we might find value in the secondary market as well.

Depending largely on how the US housing market pans out next year, HEL spreads may

widen and be more volatile, which will bring value and new opportunities to future mezzanine

SF CDOs. So stay tuned.

31 March 2006


Impact of HEQ Available Funds Caps on ABS CDO Tranches

112

Introduction Since the mid- to late 1990s, hybrid ARM loans have been growing in popularity among

mortgage loan borrowers, notably in the subprime HEQ space. The securitization rate of

this kind of product has also jumped dramatically since 2001 and taken a lot of market

share from other mortgage products. The two most popular structures are so-called “2/28”

and “3/27” ARMs. In a “2/28” ARM, the loan rate is fixed in the first 2 years and then resets

to six-month LIBOR plus a margin. (Because these are 30-year loans, the name stems

from the fact that the loan is fixed for 2 years and then floats for the remaining 28).

However, the rate after reset is subject to an initial cap, which limits the rate reset (usually

set at 3%), periodic cap rates (usually 1%-1.5% per reset), and a maximum lifetime cap

rate (usually about 600 bps higher than the initial loan rate). Hybrid ARMs provide a rate

advantage to borrowers during the fixed rate period but also expose the borrowers to the

risk of a rate increase at the expiration of the fixed period. Generally, there is a significant

increase in prepayments at the first rate reset, as many borrowers who can refinance into

a new loan do so; this is driven by the overall rate environment, credit improvement of the

borrower, and/or levels of home price appreciation.

ABS CDOs have evolved significantly over the last several years as a result of changing

market conditions, investor desires, and improving structuring technology. The differences

between latter-day ABS CDOs and ones originated at the asset class’s inception in 1999-

2000 are vast, and we’ve argued that today’s deals represent a “second-generation”

ABS/SF CDO113. Early ABS CDOs were created as a vehicle to gain diversified exposure

to many and various ABS sectors, including auto loans, credit cards, home equities,

manufacturing housing, but also esoteric assets such as aircraft securitization, franchise

loan ABS, and mutual fund fee deals. However, given the poor performance of some of

these sectors, recent ABS CDO deals have been shifting more towards residential and

commercial mortgage-backed securities. Home equity (sometimes called ”residential B&C

mortgages”) has become a much larger share of CDO collateral since 2002, concentrated

in single-A and triple-B paper.

In general, we think that this is a good thing and that the credit performance of mortgage-

related structured finance paper has been strong and is likely to remain so (again, see our

January piece and other previously referenced home equity research for more on this

topic).

Exhibit 105 shows the typical collateral mix for early-vintage ABS CDOs. HEL and RMBS

B&C only accounted for less than 10% of the pool.

And Exhibit 106 shows the collateral of ABS CDO deals of newer vintages. HEL and

RMBS B&C now take a share of 37%.

112 This section was originally written by Neil McPherson, David Yan, Rod Dubinsky and Helen Remeza,

August, 2004. 113

Please see CSFB's ABS research report "The Compelling Case for SF CDOs" (January 27, 2004)

Subordinate BBB-

rated HEQ floaters

become a large

share of CDO

collateral

31 March 2006


Exhibit 105: Collateral of ABS CDO – Early Vintage

CC

6.4%

CMBS

25.2%

MH

11.9%

SBL

1.7%

Corp

8.3%

Mutual Fund Fees

2.7%

Other

1.5%

REIT

2.9%Tax Liens

2.2% Auto

2.1%

RMBS

4.2%

CBO

11.8%

Equip

2.0%ETC

5.9%

Franchise

1.6%Future Flow

0.5%

HEL/RMBS B&C

9.0%


Exhibit 106: Collateral of ABS CDO Deals – Newer Vintage*

CBO

7.6% CC

3.4%

CMBS

15.8%

Corp

4.0%

HEL/RMBS B&C

37.1%

RMBS

10.7%REIT

6.5%

SBL

2.0%

Auto

2.1%

MH

6.7%

Other

0.8%

Mutual Fund Fees

0.6%

Structured Settlement

1.0%

ETC

1.2%Equip

0.6%

* As of August, 2004


In the last two years, we’ve seen record supply of HEQ (and triple-B HEQ bonds) matched

by equally strong demand from CDO managers in their ramp-up stage. Given the current

situation in which the bid for subordinate floaters from CDO managers is exceptionally

strong, it is important to understand the impact of AFC risk in subordinate HEQ floaters on

ABS CDO tranches. Should interest rates rise significantly in the next two to three years,

there could be a considerable impact on the cash flow and return of CDOs backed by

these bonds as a result of cap risk.

As stated earlier, available funds cap (AFC) risk refers to the situation in which the interest rate on the loans backing a subprime home equity floater is capped so that the home equity bond receives less than the rate promised. Factors influencing the AFC risk include the following:

• Coupon during initial fixed rate term.

• The periodic and lifetime rate caps of the loan.

• The margin of the home equity floating bonds (i.e., the spread over one-month LIBOR).

• Increasing prevalence of mixed pools with both hybrids and fixed rate collateral.

31 March 2006


• Deep mortgage-insurance (MI) fees netted from the loan WAC114.

• Structural credit enhancements, such as overcollateralization (OC).

• Hedging vehicles, such as cap agreements.

• Presence of IO (interest only) tranche in the deal.

Given high demand for floating rate assets and the unprecedented sharp drop in interest

rates (LIBOR) in recent years, subprime issuers can enhance the economics of

securitizations by issuing LIBOR-based floating bonds backed by a collateral pool with

some percentage of fixed rate loans – commonly 20%-30% of the pool. But this exposes

the investors of these bonds to more AFC risk, because the fixed loans are fixed for life,

and in addition, prepayment speed differentials that are different from pricing assumptions

can cause the floating/fixed mix to change significantly over time.

The AFC strike rate for a HEQ pool mixed with fixed rate and ARM loans is dependent on

the percentage and WAC of fixed rate loans in the deal as well as the relative prepayment

differential between the fixed rate and the ARM loans over the life of the deal. 115

Structural enhancements in the home equity deals, such as overcollateralization (OC),

also affect the amount of funds available for coupon payments. The higher the existing OC

level, the more funds it will generate to make coupon payments if shortfalls arise. We

should note that Moody’s and S&P recently revised their interest rate stresses used in

rating HEQ deals; the net effect is to increase OC requirements in newer HEQ deals.116

In most deals, the interest shortfall due to AFC can be carried forward and be paid by

excess cash flow. However, given that AFC shortfall payments are usually paid at the

bottom of the waterfall, the value of their offsetting the AFC risk should be significantly

discounted. The increase in OC will improve the likelihood of paying back the basis risk

shortfall.

In many deals, the trust also purchases a cap contract to mitigate the cap risk. However,

the cap contract usually covers only a limited time period, and the strike rate is often set

far out of the money.

While the AFC risk is applicable to all home equity ABS tranches, its impact is greater for

more subordinated tranches. First, the more subordinated tranches usually have longer

average lives. Thus, it is more likely to breach the strike rate, which is the same logic as a

longer-term option having a higher value, all else equal. Second, the spread or margin on

the subordinated bonds is higher than the spread on senior bonds, making it more likely to

hit the cap rate.

To simplify our analysis and concentrate on the differences in the impact of HEQ BBB

tranches on CDOs and their inherent AFC risk, we use only one bond as the entire

collateral pool (in our first three examples). This, to some extent, thus represents “worst-

case scenarios,” because in typical ABS CDOs, a portion of the collateral pool perhaps

equal in size to the HEQ portion will be in completely uncapped floaters. In addition, as

we’ll see later, the presence of a multitude of HEQ tranches in a given deal can act to

reduce the AFC risk of any one HEQ bond to the ABS CDO.

114 Weighted Average Coupon.

115 Usually defined as the net WAC (initial WAC minus servicing fee and MI premium), minus bond

spread/margin, plus payment from cap contract 116

See CSFB's Market TABS dated August 2, 2004 for a description of the S&P revision.

The mix of fixed and

floating loans

backing home

equity floaters is

one of the main

determinants of AFC

risk

While the AFC risk is

applicable to all

tranches, its impact

is greater for more

subordinated

tranches

31 March 2006


Sample ABS CDO Deal We use a hypothetical ABS CDO structure to run our analysis. Exhibit 107 shows the

detailed capital structure of the CDO and the tranche spreads.

Exhibit 107: Sample ABS CDO Deal Structure

Tranche Percentage Floating/Fixed Rating Spread

A 78% Floating AAA L+38

B 14% Floating AA L+100

C 4% Floating BBB L+315

Equity 4% Residual N/A N/A


We also make other assumptions as listed below in Exhibit 108:

Exhibit 108: Assumptions

Coupon Payment: Quarterly

Rapid/Turbo Structure:

1. All principal amortization will be used to pay down Class A notes and Class B

notes pro rata. However, in case any IC or OC test is breached, A and B notes

will revert to sequential for the rest of the transaction.

2. The Equity tranche is capped at an annual return of 15% for the life of the

transaction or until the Class C is paid off. Excess cash flows will be used to pay

down Class C notes.

Class A/B Overcollateralization Test 105%

Class C Overcollateralization Test 102%

Class A/B Interest Coverage Test 115%

Class C Interest Coverage Test 110%

Fixed-rate Voluntary Prepayment Curve Ramp up to 20 CPR from month 1 to 12 and remain constant at 20 CPR thereafter

ARM Voluntary Prepayment Curve

Ramp up to 35 CPR from month 1 to 14, stay constant at 35 CPR through month

23, jump to 70 CPR in month 24, and then ramp down to 35 until month 31 and

remain at 35 CPR thereafter

Fixed-rate Default (CDR) Curve* Zero for the first 6 months, ramps up to 3.25 CDR at Month 30, and stay at 3.25

CDR thereafter

ARM Default (CDR) Curve* Zero for the first 6 months, ramps up to 5 CDR at Month 30, and stay at 5 CDR

thereafter

Recovery in Default* Immediate recovery at 60%

Source: Credit Suisse * Only used in some of the examples, as indicated later

In essence, the Turbo structure117 in the CDO utilizes a portion of excess interest (after a

capped equity return) to amortize a Triple-B CDO tranche, thereby shortening the average

life of the Triple-B and also allowing its OC cushion to build up. The overall CDO structure

is enhanced by the Turbo as relatively expensive subordination is replaced by cheaper OC.

The pro rata paydown schedule of Class A and B notes acts to decrease the average life

of the Class B notes and build OC for both classes. However, if any coverage test is

breached, the payment schedule becomes sequential.

117 Please see CSFB's ABS research report, "Relative Value of Turbo Triple-Bs in ABS CDOs (November

26, 2002)

Turbo structure can

shorten the average

life of the Triple-B

CDO tranche and

build up its OC

cushion

31 March 2006


As we stated earlier, we assume the underlying collateral of the CDO is composed of only

one floating rate home equity BBB bond and use the principal and interest cash flows

(generated by Intex) to run through our CDO cash flow model. Because both the asset and

the liabilities are floating, we ignore any hedging issues for the CDO. There is a mismatch

between the index rates (we use one-month forward LIBOR on the home equity bond and

three-month forward LIBOR on the CDO notes). We will shock them by the same

magnitude, and therefore, we will not take this into consideration.

We list the details of the three HEQ bonds issued in 2004 that we used for our analysis in

Exhibit 109. We ran each one of them separately through our cash flow model based on the

aforementioned CDO structure. (Again, our CDO deal is backed by only one bond as

collateral.)

Exhibit 109: Details of Sample Bonds

Bond A Bond B Bond C

Moody's Rating Baa2 Baa2 Baa2

Original Balance 7,000,000 11,529,000 5,609,000

Fixed Percentage 18.9% 30.1% 30.45%

2/28 ARM Percentage

(as of all ARM loans) 72% 95% 70%

Initial Aggregate WAC (1) 7.15% 7.23% 7.99%

Fixed-Rate WAC 7.42% 7.64% 7.95%

WA Gross Margin 6.59% 6.89% 7.75%

WA Intial Rate Cap 3.00% 3.00% 3.00%

WA Periodic Rate Cap 1.17% 1.50% 1.00%

Bond Spread (over LIBOR) (2) 240 195 200

Servicing Fee (3) 0.5% 0.5% 0.5%

Strike Rate during Fixed-rate

Period (1)-(2)-(3) 4.25% 4.82% 5.49%

MI Percentage 2.7% 0% 49.67%

Initial OC Level 1.65% 2.25% 0.5%

Interest Cap Hedge? Yes

3-year Term, low strike*

No Yes

3-year Term, high strike*

Source: Credit Suisse, Intex *Detail see later

At first glance, it seems that Bond A has the highest AFC risk, Bond B second, and Bond C

the last. During the fixed rate period (say, 2 years for 2/28 ARMs), it is easy to calculate the

strike rate. Take Bond A as an example. If we subtract just the bond coupon (240 bps) and

servicing fee (0.5%) from initial WAC (7.15%), we get the strike rate at 4.25%. However, for

Bond B and Bond C, it is higher: at 4.82% and 5.49%, respectively. But, after the initial fixed

period, the strike rate also depends on the mix of ARM and fixed rate, cap contract (hedging

vehicle in the HEQ deal), and periodic cap rate and lifetime cap rate of loans, among other

things. Furthermore, the usage of MI insurance will also lower the strike rate (as MI premium

is paid from WAC). We will see that the conclusion is actually precisely the opposite: Bond A

has the lowest AFC risk, Bond B second, and Bond C the highest.

To simplify the

analysis, we use

one BBB home

equity floating bond

as the collateral and

run its cash flows

through the CDO

model

31 March 2006


Impact on CDO Deal Backed by Bond A We first assume zero CDR and that the bond will pass its delinquency trigger as well118. In

our base case, we use the forward curves and the prepayment curves aforementioned.

Because there is not much AFC shortfall by shocking the forward curve by 100 or 200

basis points (bps), we shock the curves by 300, 400, and 500 bps, respectively,119 and

reduce the prepayment speed on only the fixed rate loans by 25%, as we believe when

rates rise, prepayments will slow and the fixed rate loans will be impacted to a relatively

larger degree than the ARMs. Note shocking the forward curves in this manner is fairly

extreme; +300 bps implies that three-month LIBOR, for example, rises immediately to 5%

and to about 7% over two years.

Exhibit 110 shows the base three-month forward curve we used, as well as the curve after

a 500-bps shock.

Exhibit 110: Forward Curves

0%

2%

4%

6%

8%

10%

12%

14%

0 6 12 18 24 30 36 42 48 54 60 66 72 78 84 90 96 102 108 114 120 126

Month

Base 3-Month Forward Curve Shocked by 500 bps


Exhibit 111 shows the result of the sample CDO deal backed by Bond A. For the CDO

tranches, in the base case, both Tranche AAA and Tranche AA share the same WAL

because they are paid pro rata. Tranche BBB has a very short WAL of 3.81 as a result of the

Turbo feature. Notice all the IRR and DM120 numbers are based on a bond priced at par.

118 We could assign a delinquency level (such as 30%), but it will not change the conclusion of this report

much. Failing the delinquency trigger only changes the step-down date and the cash flows after the target step-down date (usually 3 years after origination). Further, failing a delinquency trigger does not necessarily cause loss. 119

All the curves - the six-month LIBOR on the HEL loans, the one-month LIBOR on the floating bond, and the three-month LIBOR on the CDO liability side - are shocked by the same magnitude, such as 300 bps. We start from 300 bps because there is not much shortfall if only shocked by 100 or 200 bps. 120

Discount margin, calculated based on forward curve.

31 March 2006


Exhibit 111: Result of CDO Deal Backed by Bond A – Zero CDR and with Turbo

Home Equity

Bond Tranche AAA Tranche AA Tranche BBB Equity

WAL WAL IRR DM WAL IRR DM WAL IRR DM IRR

Base Case, Forward Curve 4.23 4.14 4.49% 0.38% 4.14 5.11% 1.00% 3.81 7.37% 3.15% 19.18%

FWD+300,

Slow Fixed VCPR by 25% 4.52 3.86 7.16% 0.38% 7.34 8.63% 1.00% 2.12 9.36% 3.15% 19.30%

FWD+400,


FWD+500,



In the first stressed case (+300), there is already a coupon shortfall on the HEQ tranche

starting from the second month. However, for this bond, fortunately, all the shortfalls in this

scenario are paid back throughout the entire life of the HEQ bond. This is the case for the

second stressed scenario (+400) as well. There are three main reasons for the low AFC risk:

1. Relatively low percentage of fixed rate loans (19% initially).

2. A cap contract (three-year term) with high initial notional and low strike rate, which

makes up most of the AFC shortfall. The cap is in-the-money early on even in the

300-bps stress scenario.

3. Excess spread, which also covers the rest of the shortfall.

So, when shortfalls can all get paid back (which is equivalent to no shortfalls)121, it turns

out that increasing the interest rate raises the coupon rate on the HEL bond and thus

generates more excess flows to the CDO. As a result, the BBB tranche gets paid more in

interest and paid off faster, thanks to the Turbo feature, as shown by its short WAL and

high IRR numbers.

However, in the last stressed scenario (+500), after about five years the shortfalls are too

high to be paid back (remember there are shortfalls in the more senior tranches of the

HEQ deal and they need to be paid down earlier). Thanks to the Turbo feature, most of the

BBB tranche has already been paid down, the WAL only extended slightly from 2.13 to

2.72. Only the equity tranche got hurt as its IRR dropped to 14%.

Overall, in all scenarios, the DM always holds at the original floating spread for each

tranche.

Note that in the stressed scenarios, the WAL of Tranche AAA is much shorter than the

WAL of Tranche AA. This is because the IC test failed once, so they switched to

sequential pay-down.

Some people might argue that as there is default or credit loss, the excess cash flow will

be reduced so that the shortfall will not be paid back. We will assume certain non-zero

CDR curves in our next example.

What if we took away the Turbo feature? We would expect the WAL of BBB to increase

significantly (even in the base case) because now it will be paid off after AAA and AA. This

is exactly the case as shown in Exhibit 112, where we compare only the BBB and equity

tranches with and without Turbo.

121 In Bond A's case, all the shortfalls are actually paid back in the same period as the shortfall occurs.

Low percentage of

fixed rate collateral

and low-strike cap

contract make the

AFC shortfalls of

Bond A literally zero

under all stressed

scenarios used

31 March 2006


Exhibit 112: Result of CDO Deal Backed by Bond A – Zero CDR BBB Equity

Without Turbo With Turbo Without Turbo With Turbo

WAL IRR DM WAL IRR DM IRR IRR

Base Case, Forward Curve 7.09 7.88% 3.15% 3.81 7.37% 3.15% 23.11% 19.18%

FWD+300,

Slow Fixed VCPR by 25% 7.88 10.46% 3.15% 2.12 9.36% 3.15% 26.36% 19.30%

FWD+400,


FWD+500,



A more important observation is that the WAL of the CDO BBB in the stressed scenarios is

now longer than the WAL in the base case when the Turbo feature is absent (about 0.8

years longer). When the HEQ prepayment is slower, there is less principal cash flow; thus,

the balances of AAA and AA tranche are paid down more slowly. With the Turbo removed,

the CDO’s BBB tranche is pushed far back in the line to receive cash flows. In addition,

the DM on the BBB tranche is reduced when Turbo is removed. On the flip side, the equity

tranche is better off (higher returns).

Usually, in real CDO deals, the difference in WAL between Turbo and no-Turbo deals is

about three years. The reasons that the difference here is so dramatic (in the +300-bps

case, it is about 5.6 years) are as follows:

1. We use only one bond as collateral, which makes it extremely sensitive to the

structural assumptions we use. In reality, there will be not only more home equity

bonds but also other asset types, such as credit card receivables, auto deals,

CMBS, etc.

2. The Turbo BBB accounts for a relatively small share of the whole CDO deal (only

4%), which makes it even more sensitive – i.e., it does not take much cash flow to

pay it down if funds are available.

3. In our deal, there is also no reinvestment period, which in a real deal acts to

extend the average life of a Turbo BBB tranche; non-Turbo BBB tranches are

rarely longer than ten years even with a revolving period.

Overall, the ABS CDO deal backed by Bond A is not exposed to significant AFC risk. This

is again due to the following three factors: a relatively low percentage of fixed loans, the

absence of MI coverage, and a cap contract with high notional and low strike rate. As a

result, the high margin on the bond can be fully enjoyed by investors.

Impact on CDO Deal Backed by Bond B We make the same assumptions regarding Bond B. The biggest differences between

Bond A and B are that B has a larger percentage of fixed rate loans (30%) and that it does

not have a cap contract in the deal. As ARMs typically pay down faster than fixed rate

loans and the AFC strike rate migrates towards the coupon on the fixed rate loans, this is

especially problematic for bond B, because there is no cap contract present. Bonds with a

higher percentage of fixed rate loans have more AFC risk than those with lower

percentage of fixed rate loans, everything else equal.

Exhibit 113 shows the results of the CDO deal backed by Bond B. Notice all the IRR and

DM numbers are again based on a bond price at par.

Bond B has more

fixed rate collateral

and does not have

cap contract in the

deal

31 March 2006


Exhibit 113: Result of CDO Deal Backed by Bond B – Zero CDR and with Turbo

HEL Bond Tranche AAA Tranche AA Tranche BBB Equity



FWD+300,


FWD+400,


FWD+500,

Slow Fixed VCPR by 25% 5.05 4.12 9.11% 0.38% 9.28 10.78% 1.00% 9.55 3.82% N/A* N/A*

Source: Credit Suisse, Intex *N/A means a large negative number

The first difference we notice is that even in the base case, the BBB tranche backed by

Bond B has a longer WAL than that backed by Bond A (7.26 vs. 3.81). This has to do with

another difference between Bond A and B – A has a much higher margin than B (240

versus 195, a 45-bps difference). When the AFC is not in effect, higher margin means

more interest cash flow and excess flow to the CDO, especially the BBB tranche. The

Turbo feature just makes the difference more dramatic. When there is not enough excess

flow, the CDO BBB could be completely shut out from any payment. In the case of Bond B,

there is a period of a little over four years during which it does not receive any principal. It

gets paid off after the AAA and AA tranches are paid off. Remember that before step-down,

the bond does not receive any principal, so the only thing that can be used to pay the BBB

tranche is the interest cash flow in excess of the equity cap (15%).

In both the first (+300 bps) and second (+400 bps) stressed scenarios, the WAL of the

CDO tranche BBB shortens significantly. As we take a closer look at the AFC shortfalls

and whether and when they get paid back, here is what we find: In the first stressed (+300

bps) scenario, all the AFC shortfalls are paid back in the first 86 months, and only twice is

the payback made with a 1-month lag. In the second stressed (+400 bps) scenario,

although the lag could be up to 12 months, all the shortfalls are paid off by the first 82

months. Consequently, as we discussed earlier, when shortfalls can be paid back (such as

the +300 and +400 cases here), there is actually more excess flow to the CDO, and with

Turbo, the BBB is paid off faster. In addition, all the IRR and DM numbers hold up well.

However, it is a totally different case when the AFC shortfall cannot be paid back. In the third

stressed scenario, where the forward curve is shocked by (an unrealistic) 500 bps, the

shortfalls are too high to be paid back – as a matter of fact, no shortfalls are ever paid back

in this scenario. Remember that the AFC shortfall can only be paid back at the bottom of the

waterfall and from senior tranche to junior tranche. In this case, there is nothing left to pay

the shortfalls on the Triple-B bond (raising interest rates also increases the AFC shortfall for

more senior bonds). As a result, the BBB tranche is extended to a WAL of 9.55 and has a

much lower IRR of 3.82%. DM also turns to a large negative number.

Obviously, Bond B is worse than Bond A in terms of AFC risk protection. Remember that

when the CDO is backed by Bond A, the impact of AFC on BBB is still very limited when

interest rates are shocked by 500 bps! (Both the IRR and DM hold well.) The main factors

hurting Bond B include a high percentage of fixed rate loans and lack of a hedging vehicle,

such as a cap contract.

It is even worse for the equity tranche. When the rate is shocked by 500 bps, the equity

tranche gets almost nothing. Any cash flows that could have gone to the equity are

diverted to make up any interest shortfall on the more senior tranches.

A CDO deal backed

by Bond B has

greater AFC risk

than the one backed

by Bond A

31 March 2006


Another interesting observation is that the WAL of AAA and AA tranches are no longer the

same in the second stressed (+400-bps) scenario. Based on our set-up in this article, the

OC test in the CDO never fails, but the IC test could fail122. As the IC test fails, the AAA

and AA are paid off sequentially rather than pro rata; thus, AAA’s WAL shortens while

AA’s lengthens. Exhibit 114 shows the IC ratio for AA tranche backed by Bond B in the

first 24 periods (quarters). It is clear that the IC test fails early in Scenario 3 and 4. The

spike in Scenario 3 around the eighth period is caused by a large payback in the AFC

shortfall.

Exhibit 114: IC Ratio of AA Tranche Backed by Bond B

0%

50%

100%

150%

200%

250%

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24

Period

Ratio

AA IC_Base Case AA IC_+300 bps AA IC_+400 bps AA IC_+500 bps

Fail

Pass

AA IC Trigger: 115%


We also check the results without the Turbo feature, as shown in Exhibit 115. The

conclusions are very similar to those in the case of Bond A. One difference is that, as in

the case with Turbo, the IRR in the 500-bps-shock scenario is much worse than that of

Bond A and the DM turns negative. Another interesting observation is that even in the

+400 bps scenario, the DM drops to 1.39% without Turbo.

There are two benefits of the Turbo structure for BBB investors here:

1. The performance is a little bit better with Turbo in the +500 scenario: shorter WAL

and higher IRR and DM.

2. Without serious AFC shortfall (+400 scenarios), a Turbo structure will significantly

improve the performance: much shorter WAL and higher DM.

122 Normally, the OC test will be triggered before the IC test is triggered.

AFC shortfall could

cause the IC test to

fail and thus change

the cash flow of the

whole capital

structure

Turbo feature

benefits the BBB

investors

significantly

31 March 2006


Exhibit 115: Result of CDO Deal Backed by Bond B – Zero CDR

BBB Equity




FWD+300,


FWD+400,


FWD+500,

Slow Fixed VCPR by 25% 10.49 2.52% N/A* 9.55 3.82% N/A* N/A* N/A*

Source: Credit Suisse, Intex * N/A means a large negative number

Now let us apply our CDR curve, as shown in Exhibit 108, to our analysis. Increasing CDR

will increase loss and eventually cut into OC. It could postpone the step-down date (if the

HEQ deal’s OC level does not meet the target) and thus change cash flows to the HEQ

bond and the CDO. Or, it could exhaust the entire OC cushion and cause principal loss to

the most junior HEQ tranche first and continue to work from the bottom up. Higher CDR

and credit loss will lower the excess spread available (by reducing the performing balance)

for making up any AFC shortfall, and thus, more shortfall will accumulate. So, a higher

default rate could have an effect on the AFC shortfall. Exhibit 116 shows the results with

Turbo. Comparing the numbers in Exhibit 116 with those in Exhibit 113, we have the

following observations:

1. Overall, the results for AAA, AA and BBB tranches are very similar with and

without CDR. CDR curves normally start from very low levels and then ramp up to

a certain level around 30-36 months out and then level off. A CDR curve with this

kind of profile will generate moderate loss early on and thus has minimum impact

on the short AAA, AA and BBB tranche. However, under an extremely stressed

scenario, such as forward curve plus 500 bps, the shortfall is so high that there is

not much excess spread left to cover the loss. Eventually, the OC drops too low

and the principal payback on the Triple-B CDO bond is further delayed; thus, the

WAL of the BBB tranche jumps from 9.55 to 13.82.

2. In the stressed scenarios, the return on the equity tranche turns still lower. With

Turbo, the cash flow to the equity tranche tends to be back-loaded. As discussed

above, a higher default rate will increase the AFC shortfall and decrease cash

flows to the equity tranche.

Our CDR curves result in a lifetime cumulative loss of around 4.2% for fixed-rate and 2.5%

for ARMs in the base case, given our prepayment and recovery assumptions123. While

these CDR curves are relatively “light,” we should note that our interest rate stresses are

so “heavy” that the probability for them to happen is extremely slim; in other words, excess

spread is squeezed so much given these interest rate stresses that higher CDRs cannot

be sustained.

123 Although our assumed CDR curve is higher for ARM than for fixed rate, the cumulative loss rate is

higher for fixed rate since the prepayment speed of fixed rate is slower.

Higher default/loss

will decrease excess

spread and thus

increase AFC

shortfall

31 March 2006


Exhibit 116: Result of CDO Deal Backed by Bond B – With CDR Curve and with Turbo




FWD+300,


FWD+400,


FWD+500,


*N/A means a large negative number

Source: Credit Suisse, Intex,

Impact on CDO Deal Backed by Bond C Compared with Bond A and B, Bond C not only has a higher percentage in fixed rate loans

(relative to Bond A) but a higher percentage of loans covered by mortgage insurance (MI)

as well. Because the MI premium is taken out of the WAC, a larger share of MI covered

loans will increase the AFC risk, even for a high WAC bond, such as Bond C (7.99%).

Another disadvantage that Bond C has is that its OC level is much lower than the OC level

of Bond A and Bond B (we think this is probably because it has a high MI coverage). It is

only 0.5% versus 1.65% for Bond A and 2.25% for Bond B. As mentioned earlier, a low

OC level will exacerbate the AFC risk. Last, like most HEQ floating bonds, Bond C’s

collateral has a periodic cap of 1%, which is lower than Bond A’s (1.17%) and Bond B’s

(1.5%)124, which further limits the interest collection from the HEQ loans and increases the

AFC risk.

Exhibit 117 shows the results of the same sample CDO deal backed by Bond C. And

Exhibit 118 shows the results without Turbo. We assume zero CDR here again. Overall,

the conclusions are very similar between Bond B and Bond C. We believe that the much

higher WAC of Bond C and the cap contract within the deal offset its higher percentage of

MI. Given that, we still prefer Bond B a little bit more than Bond C. Exhibit 119 shows the

IC ratio of AA tranche backed by both bonds in the +300-bps scenario. The lines are pretty

much on top of each other early on before they reach around period 20, where the two

lines start to diverge. Around period 22, the AA backed by Bond C starts to fail the test

(that is why the WAL of AA in the Bond C scenario is a little bit longer than that of AAA).

Exhibit 117: Result of CDO Deal Backed by Bond C – with Zero CDR and with Turbo




FWD+300,


FWD+400,


FWD+500,


Source: Credit Suisse, Intex, *N/A means a large negative number

124 Bond A and Bond B both have a higher-than-average periodic cap.

High percentage of

fixed rate loans and

MI coverage and low

OC level make the

CDO deal backed by

Bond C exposed to

even higher AFC risk

31 March 2006


Exhibit 118: Result of CDO Deal Backed by Bond C – with Zero CDR BBB Equity




FWD+300,


FWD+400,


FWD+500,

Slow Fixed VCPR by 25% 8.08 -0.16% N/A* 7.14 0.51% N/A* N/A N/A

Source: Credit Suisse, Intex, *N/A means a large negative number

Exhibit 119: IC Ratio of AA Tranche Backed by Bond B and C

+300-bps Scenario

0%

50%

100%

150%

200%

250%

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24

Period

Ratio

AA IC_Bond B AA IC_Bond C

AA IC Trigger: 115%

Fail

Pass


The last thing we would like to point out is that the cap contract associated with Bond C

has a much higher strike rate (well above 6%, and this varies depending on time and

group125) than the strike rate of the cap contract associated with Bond A. Given that the

deal also has a higher percentage in fixed rate loans and almost 50% MI coverage, the

protection provided by its cap agreement could be limited in more stressed scenarios.

Putting All Three Bonds Together Some good news is that, in reality, ABS CDOs are backed by different asset classes and,

within the same asset class, different bonds with various features. So we created a

“portfolio” of these three bonds, equally weighted, and used this “diversified” pool to back

the CDO deal. Not surprisingly, the results are less extreme.

Exhibit 120 shows the results with zero CDR and Turbo. Although the WAL of the BBB

tranche still gets extended to 8.57 in the third stressed scenario (+500), the return stays high

at 12.7% and the DM only drops to 2.57%. Compared with the results of Bond B and Bond C,

where the WAL gets extended and IRR and DM are reduced to very low levels, it is clearly

better now.

125 For example, in September 2004, the strike rate is 6.6%, and in October 2004, it is 6.82% for Group 1

loans, while 6.31% and 6.52% for Group II loans. Also, the notional of the cap decreases over time.

31 March 2006


Exhibit 120: Result of CDO Deal Backed by Three Bonds – Zero CDR and with Turbo




FWD+300,


FWD+400,


FWD+500,

Slow Fixed VCPR by 25% 4.89 3.95 9.03% 0.38% 7.33 10.50% 1.00% 8.57 12.71% 2.57% N/A*


Remember that WAL is a measurement of how fast the principal gets paid back, while IRR

is determined by not only the principal payback but also how much interest can be

received. In the +500-bps scenario, because of AFC shortfall, there is not much cash flow

left for paying interest on BBB tranche after paying interest on the AAA and AA tranches in

the first seven years or so, and the entire principal cash flow goes to AAA (from a switch to

sequential pay due to IC test failure). As a result, not much gets paid on BBB’s principal –

not until both AAA and AA tranches are paid off.

On the IRR front, thanks to the interest cash flow from Bond A, the BBB tranche gets paid

in interest after seven years (no need to pay interest on AAA and AA tranches because

they have been paid down, yet nothing is left for equity tranche); thus, its IRR gets boosted

to 12.7%126. Remember again that, when the CDO is backed by Bond B or C, the BBB

tranche does not get much from interest payment in the +500-bps scenario.

So, as can be seen, the presence of several bonds with different features should act to

reduce any extreme impact AFC risk might cause on ABS CDO deals.

Conclusion In this article, we investigated the impact of available funds cap risk of home equity bonds

on ABS CDOs with regard to average life, return, and the IC compliance test.

As can be seen, AFC risk is quite different depending on the home equity collateral bond

structural and collateral features. In addition, the impact of AFC on different tranches of

the CDO could vary and can also depend on the structure of CDO (such as deals

with/without a Turbo). Our major findings/conclusions are summarized as follows:

From a macro perspective, we believe interest rates would need to rise dramatically before

sizable AFC risk affects the cash flows of ABS CDOs. Based on our analysis, three-month

LIBOR would have to jump to 9% in two years (as of now, it stands at around 1.7%). This,

of course, represents a very extreme scenario and is probably not very likely to happen.

Moreover, hopefully, the economy will improve as rates go up, and thus, defaults on the

home equity loans will go down.

From a micro perspective, investors, as well as CDO managers, need to be aware of the

drivers of AFC risk. To mitigate or avoid AFC risk, CDO managers should try to pick

floating bonds with the following features to the extent possible: a relatively low

percentage of fixed rate loans, a low percentage of MI coverage, higher available OC level,

and effective hedging vehicles (such as a cap contract). In addition, for BBB-tranche CDO

investors, a Turbo structure will further protect them from AFC risk and, when AFC

shortfall is absent, improve the performance significantly. Finally, by investing in different

bonds, the diversity effect will likely prevent the extreme scenarios from happening and

thus avoid large losses from AFC risk. We hope this piece is useful to investors as a

framework for understanding the extent that ABS CDOs are exposed to AFC risk.

126 Notice that even when the CDO is backed by Bond A only, in the +500-bps scenario, the IRR for BBB is

11.55% (lower than 12.71% here). That is because, when only Bond A is used, the BBB is paid down fast (WAL only at 2.72) and the rest of the cash flow goes to equity. However, with three bonds mixed, the WAL is extended to 8.57, and thus, some of the cash flow goes to pay BBB's interest rather than it all going to equity.

31 March 2006


Chapter 2. Collateralized Loan

Obligations (CLOs)

31 March 2006


Calling Attention to CDO Calls127

Overview CDO calls have recently become one of the hottest topics in the secondary CDO market.

We estimate up to $43 billion of CDOs will exit their non-call period over the next 12

months. This is based on the typical non-call period of 3-4 years and recognizes the

majority of CDOs called to date are emerging market (EM) and high yield (HY) CDOs.128

With many CDOs trading at a premium, it is imperative that investors understand the

drivers of the decision to call.

We provide an overview of CDO calls in four points:

1) Call Provisions: The Nitty-Gritty. We present the call provision found in most

deals and the basic requirements for a deal to be callable.

2) Call Rationale. A discussion of the primary drivers of CDO calls and whether

they make economic sense.

3) Candidates for Call. Characteristics of CDOs which may be good call

candidates from the perspective of the equity holder.

4) Blocked Call: Deterrents to Calls. Factors that may deter the callability of a deal.

We provide a list of known called CDOs at the end of this report.

Call Provisions: The Nitty-Gritty A typical CDO carries a set of optional redemption provisions (Article IX in most

indentures). Based on the Optional Redemption provision, at the direction of a super-

majority (two-thirds), or in some cases a majority (one-half), of the equity holders, a deal

may be called following the end of the non-call period, usually about 3-4 years from the

closing date.

In order for a deal to be called, liquidation proceeds must be sufficient to cover the

following liabilities and expenses:

• Hedge termination/unwind fees;

• Aggregate outstanding principal;

• Accrued, unpaid and deferred interest (PIK-able tranches);

• Make-Whole provisions. Some deals require an optional redemption premium for fixed-

rate tranches. See below for a detailed discussion;

• Administrative and other fees & expenses;

• Asset manager and advisory fees. Some deals also include the present value of forgone

management fees (as a result of the early termination) for a predetermined period.


128 HY CDO includes both CBOs and CLOs.

31 March 2006


The Make-Whole provision merits additional insight. While floating-rate tranches are

typically callable at par, certain deals with fixed-rate tranches may be due a premium

based on the present value of remaining fixed-rate tranche cash flows, discounted at a

predetermined rate. Determining the remaining cash flows varies from deal to deal, but

the most popular approaches include:

1. Taking the coupon payments at each remaining payment date plus the principal

(equal to the remaining balance of the tranche) paid back on the expiration date of the make-whole premium;

2. Dividing the outstanding principal amount evenly among the remaining payment dates between the redemption and the maturity date or an expiration date (usually consistent with the expected average life of the tranche at closing), plus the interest on the notes for each remaining payment date;

3. Utilizing a predefined principal amortization and interest schedule table.

The actual discount rate is typically calculated as a spread over Treasuries. In general,

the Treasury rate is determined by interpolating the Treasury yields, reported on a specific

date before the redemption date, to the remaining life of the tranche. While the discount

spread also differs across deals, many CDOs calculate the spread as half the

liability spread over Treasuries of the tranche at issuance.129

Call Rationale A call is economically viable if the present value of the call proceeds exceeds the

present value of expected future equity cash flows without the call, i.e. :

Present Value of Call Proceeds > Present Value of Cash Flows if No Call (Equation 1)

Many people prefer to use IRR instead of present value. In terms of IRR, Equation 1

equivalent is:

IRR of Reinvesting Call Proceeds > IRR of Future CFs by Investing Call

Proceeds in the CDO (Equation 2)

For example, assume the net proceeds from the call equal $10 million and by investing this

elsewhere the equity holders can earn 12% IRR. To call the deal, the IRR on the future cash

flows from the CDO has to be below 12%. Alternatively, using the present value methodology,

the present value of future cash flows discounted at 12% has to be below $10 million for the

call to be economically rational. The equity holder can also use a scenario analysis to see the

profile of both approaches based on the cash flows generated under different assumptions,

such as default and prepayment rates, to decide whether to call the deal.

We discuss two primary drivers for CDO calls: asset spread compression and liability

spread compression.

Asset Spread Compression

When asset spreads tighten, asset prices appreciate, motivating the equity holder to

liquidate the assets in order to lock in gains. A strong credit environment over the last two

years has driven spreads in for many collateral markets, particularly the emerging market

(EM) and high yield bond (HY) sectors. Not surprisingly, this has fueled the up-tick in EM

and HY CBO calls in recent years, as shown in Exhibit 121. The shadowed bars indicate

the number of deals called at each point in time.

129 Please note that fixed-rate tranches are usually priced at a spread over swap.

31 March 2006


The price appreciation in leveraged loans is relatively limited, as loans are typically floating

rate and callable at par. However, the fact that many CLOs could have other asset types,

such as HY bonds (usually up to 10%), included in the collateral can raise the price further.

Even if the collateral is not trading at a significant premium, it is still economically possible

for a deal to be called, although it may not be on the first redemption date.

Exhibit 121: Asset Spread Compression vs. EM/HY CBO Calls

0

100

200

300

400

500

600

700

Oct-00

Jan-01

Apr-01

Jul-01

Oct-01

Jan-02

Apr-02

Jul-02

Oct-02

Jan-03

Apr-03

Jul-03

Oct-03

Jan-04

Apr-04

Jul-04

Oct-04

Jan-05

Apr-05

Asset

Sp

read

(b

ps)

0

1EM/HY CBOs Called *BB US HY Bonds (CSFB HY Index, sw aps)BB CSFB SBI (sw aps)BBB CSFB SBI (sw aps)

* Each vertical bar represents one EM/HY CBO called, i.e. the thicker the bar, the more deals called at that time.

Source: Credit Suisse, based on CREDIT SUISSE’s Sovereign Bond Index and High Yield Bond Index

Liability Spread Compression

We think asset spread compression is closely related to CDO liability spread compression.

The latter is mainly a result of an improved credit environment, advancements in

structuring technology, a maturing secondary CDO market, and most importantly,

increasing investor acceptance and product demand.

The compression in liability spreads may also potentially trigger CDOs to be called and

refinanced into new deals, especially for HY CLOs. We believe the refinancing is usually

initiated by CDO managers or issuers. If the managers don’t have enough share of equity

holdings, they will either recommend the equity holders call or purchase the necessary

shares from them.130 Equity holders should base their decision on Equations 1 or 2 shown

above. The cost savings may boost the potential cash flows to equity holders, although

this also depends on the return from the asset side of the new deal.

Other factors driving HY CLO calls include both limited reinvestment options because of

collateral eligibility criteria such as maturity restrictions, and limited availability of such

collateral at cost-effective prices. Moreover, the surge in leveraged loan refinancings

during the low rate environment of 2003 and 2004 left many HY CLOs with large cash

positions, which diminishes equity returns and increases the incentive to call. Exhibit 122

shows CLO call activity versus the liability spread compression.

130 For some older CLO deals, it only takes the majority share of equities to call the deal.

31 March 2006


Exhibit 122: Liability Spread Compression vs. HY CLO Calls

0

50

100

150

200

250

300

350

Aug-01

Oct-01

Dec-01

Feb-02

Apr-02

Jun-02

Aug-02

Oct-02

Dec-02

Feb-03

Apr-03

Jun-03

Aug-03

Oct-03

Dec-03

Feb-04

Apr-04

Jun-04

Aug-04

Oct-04

Dec-04

Feb-05

Apr-05

Lia

bil

ity S

pre

ad

(b

ps)

0

1HY CLOs Called *

AAA HY CLO

AA HY CLOA HY CLO

BBB HY CLO

* Each vertical bar represents one HY CLO called, i.e. the thicker the bar, the more deals called at that time.


Candidates for Call In addition to the aforementioned rationales for calls, several CDO characteristics signal

call candidates from the perspective of the equity holder. These include:

Deals at the End of or After the Reinvestment Period

As deals begin amortizing following the end of the reinvestment period (typically 5-7 years

after closing), cash flow is diverted to pay down the tranches beginning with the cheapest,

and most senior, notes. As a result, the all-in funding cost of the deal rises, reducing the

equity return. Equity holders, therefore, have greater incentive to call the deal.

Deals with Limited Reinvestment Options

Many deals were called even before the end of the reinvestment period (see Exhibit 123).

For CDOs with limited reinvestment options, the upside potential of trading gains and the

ability to take advantage of market opportunities is capped. Collateral eligibility criteria

such as average life, combined with the significant jump in secondary prices, limit the

manager’s investment options, which results in a build-up of large cash positions (from

principal proceeds and prepayments) and decreasing equity returns. Equity holders may

find more value in calling the deal, shifting their interest to other alternative investments.

Deals with Failed Coverage Tests

At the first sign of a failed coverage test, cash flow is diverted away from the equity

tranche to pay down senior notes. Equity holders must decide whether the tests can be

cured so that cash flows can be resumed, or instead there may be further future losses, in

which case calling the deal may result in a higher return.

31 March 2006


Blocked Call: Deterrents to Calls Several factors may deter the callability of a deal even though the call might make

economic sense. These include:

Identifying the Equity Holders

As mentioned, most CDOs require a super-majority (two-thirds) vote from the equity

holders to execute the optional redemption. The problem therein lies not only in getting

the consensus of two-thirds of the holders to call the deal, but also in identifying the equity

investors. For older deals, although the equity is less widely distributed, obtaining an

investor list is more difficult as the trustee is not required to disclose the list. In addition,

the equity may have changed hands several times during its lifetime. For more recent

deals, the difficulty lies in much wider equity distributions, although many trustees are now

required to disclose the holder list to requesting equity investors.

Nature of the Equity Holders

Whether the optional redemption ultimately gets exercised is dependent on the investment

nature of the equity holder. While total-return investors are generally in favor of

monetizing call potential, buy-and-hold investors, which hold most of the outstanding

equity shares (especially in older deals), are more reluctant to sell. Even if a call is

economically viable per our definition, the equity holder may still realize an immediate loss

by executing the call, which may be unacceptable to many buy-and-hold investors.

The Redemption Process

Furthermore, the redemption process is far from trivial. Following a thorough ratification

process to determine whether the call is feasible, proof that the collateral liquidation would

be sufficient to cover the liabilities and expenses must be provided to the trustee in advance

of the sale. This often takes the form of binding agreements with a highly rated counterparty

or bid-side quotes with some haircut based on the amount of time to the redemption date.

The process may require a significant amount of time, during which the market may move.

For the typical buy-and-hold investor, the ends may not justify the means.

Refinance Realities

While the rally in CDO liabilities has motivated redemptions and subsequent deal

refinancings, the reality is that several barriers hinder the ability to refinance. The

collateral in CDOs past the non-call period may be trading at a premium in the secondary

market, making it potentially difficult to roll into a new transaction. Other eligibility

concerns include: credit impaired and credit deteriorated assets, which in many seasoned

CDOs is likely; minimum ratings criteria; and average life considerations.

31 March 2006


Exhibit 123: CDOs Called

Deal Name Manager

Deal Size

($mm)

Price

Date

Non-Call

(months)

Reinvest

(months) Call Date**

HY CBO

AmVestors CBO Trust I Salomon Brothers Asset Mgmt $254 12/5/96 60 60 7/1/03

Astron CBO Orix USA $548 6/4/98 18 18 2/1/04

Cedar Lake CBO Ltd. TCW Asst Mgmt $150 7/31/02 --- 12 9/30/04

Dresdner RCM Caywood Scholl CBO I, Ltd. Dresdner/Caywood $296 10/14/99 48 43 5/2/05

IBEX Financial II Phoenix Investment Counsel, Inc. $980 8/25/98 --- --- 10/4/04

MassMutual/Darby CBO LLC DL. Babson (Mass Mutual) $482 12/23/97 61 61 5/16/05

Polar Funding I Ltd. ING Ghent Asset Mgmt LLC $300 11/2/01 12 static 4/1/04

Robeco CBO I, Ltd. Robeco NV $299 6/9/00 36 60 12/30/04

San Joaquin HY CBO I Pacific Investment Mgmt Company $240 9/28/01 36 60 11/2/04***

Topsail CBO, Ltd. ING Ghent Asset Mgmt LLC $200 3/30/01 48 48 4/23/05***

HY CLO

AMMC CDO I, Ltd. American Money Mgmt Corp. $400 11/30/99 61 61 1/26/05

Campobello Master Trust, Series 1999-1 Bank of Nova Scotia $2,658 4/26/99 --- --- 5/1/03

Commercial Loan Funding Trust I Lehman Brothers CP Inc. $823 8/20/97 --- --- 6/1/03

Eaton Vance CDO IV, Ltd. Eaton Vance Mgmt $280 3/14/01 24 60 4/1/04***

ELC (Cayman) Ltd. 1999-3 DL. Babson (Mass Mutual) $452 12/9/99 48 60 4/14/05

ELC (Cayman) Ltd. 2000-1 DL. Babson (Mass Mutual) $512 6/13/00 36 60 4/11/05***

ELC (Cayman), Ltd. DL. Babson (Mass Mutual) $431 12/22/98 37 58 1/20/05

Great Point CLO 1999-1 Ltd Sankaty Advisors $409 5/26/99 63 60 9/20/04

Harch CLO I, Ltd. Harch Capital Mgmt $425 3/10/00 38 60 3/22/05

Lakeshore Commercial Loan Master Trust I Bank of Montreal $3,051 7/17/98 --- --- 11/1/03

Northwoods Capital Ltd Angelo Gordon $475 1/28/99 48 72 9/13/04***

Pacifica Partners I, LLP Imperial Capital Mgmt./Caywood Scholl $500 8/27/98 --- --- 6/1/04

Sequils I TCW Asst Mgmt $713 4/1/99 36 84 6/1/04***

SEQUILS IV TCW Asst Mgmt $500 4/28/00 36 60 5/24/04***

Van Kampen CLO I, Ltd. Van Kampen $1,276 10/8/97 36 36 4/8/05

EM CBO

Alliance Investments, Ltd. Alliance Capital Mgmt $388 11/12/97 --- --- 5/1/04

Atlas CDO, Ltd. Ashmore Investment Mgmt $170 3/12/01 48 60 4/15/04

Augusta Funding 1997-B Bear Stearns Asset Mgmt $282 4/8/97 48 96 10/10/03***

EM CDO I * N/A N/A N/A N/A N/A 4/21/03

EM CDO II * N/A N/A N/A N/A N/A 9/3/03

Global Funding Ltd A Bear Stearns Asset Mgmt $301 1/7/98 24 96 1/8/04***

Global Funding Ltd C Bear Stearns Asset Mgmt $269 4/21/98 24 96 10/23/03***

Global Sovereign CBO, Ltd Bear Stearns Asset Mgmt $104 4/8/99 24 60 10/8/03***

ML CBO XVII Series 1998-Carlson-1 Carlson Mgmt. (Jersey) Ltd. $173 7/16/98 36 60 7/23/03

New Alliance Global CDO, Ltd. Alliance Capital Mgmt $250 4/25/01 35 59 4/12/05***

One World Global Sovereign CBO Ltd. One World Investments $198 7/19/01 36 60 7/26/04***

OUB Sovereign Emerging Markets CBO I Ltd. OUB Asset Mgmt $242 6/30/98 24 96 6/30/03***

Phoenix Global Sovereign CBO, Ltd. Phoenix Investment Counsel, Inc. $250 8/10/00 41 41 2/10/04

TCW GEM IV, Ltd TCW Asst Mgmt $231 1/22/99 24 96 7/22/03***

CRE CDO

Mach One CDO 2000-1 Bank One $310 5/15/00 36 static 6/29/04

Pinstripe I CDO, Ltd. Alliance Capital Mgmt $484 3/16/01 120 36 11/2/04

* Confidential Transaction

** In some cases, the ratings withdrawal date

*** Deals called before reinvestment period

Source: Credit Suisse, S&P, Moody’s, Intex, deal documents

31 March 2006


When’s the Best Time to Call? Optimal Timing of CDO Calls and Relative Values

131

There has been much discussion in the market on CDO calls or optional redemptions.

Notwithstanding all the non-economic factors preventing a CDO from being called, it is

important for CDO investors to understand the nuances in assessing CDO calls from an

economic perspective. Here we provide a framework for assessing CDO calls, as well as

some discussion on relative value.

The economic rationale for a CDO to be called is discussed in the previous section

(Equations 1 and 2).

While simple in principle, there are many factors impacting the value on both sides of the

equations, including the following:

1. Underlying Collateral Factors:

1) Prepayment speed/amortization rate of the underlying collateral

2) Default rate of the underlying collateral

3) Recovery rate after default

2. Market Factors:

1) The market price of the collateral at the potential call dates

2) The tightening of the liability spreads of CDOs, especially for CLOs

3) The IRR on alternative investments available to equity holders at the time

of call

3. Structural Factors:

1) How many fixed tranches in a CDO and the Make-Whole Provision

2) The hedging agreement embedded in a CDO

3) The management fee and/or incentive fees in a CDO

In our analysis, we use an actual HY CLO deal as an example. Exhibit 124 shows the

detailed information, a 2001 vintage deal with one year remaining to the end of the non-

call period. Exhibit 124 also shows the base-case prepayment, default and recovery

assumptions. By running these assumptions through Intex and using a forward LIBOR

curve, we generate cash flows for each tranche for import into our call model.

Make-whole premium on fixed tranches could be significant

An important issue is to quantify the make-whole premium of fixed tranches based on the

Make-Whole Provision specified in the indenture. Usually there is an expiration date for

the make-whole premium, before which the redemption price is the present value of the

remaining schedule of payments of principal and interest, assuming the entire remaining

outstanding principal will be paid at par on the expiration date. After the expiration date,

the fixed tranche could be called at par. Exhibit 124 also shows the expiration dates of

each fixed tranche. The make-whole premium could be calculated as:

(PV of a Par Bond (discounted at Treasury plus make-whole spread,

with a maturity from now to the expiration date) – Par)/Par


31 March 2006


Using tranche D2 as an example, the last column in Exhibit 125 shows the results. The

make-whole spread for this tranche is 371 bps based on the indenture. Given the

remaining maturity to the expiration date of 5.75 years (if called on the first redemption

date) and the corresponding Treasury rate of 3.88%, the discount rate to use is 7.59%,

much lower than the coupon rate of 12.05%. As a result, the premium needed to pay down

this tranche is a whopping 20.68%! So, if the remaining balance of D2 is $11,800,000, the

equity holders will have to pay $14,239,972 to the D2 holders if they want to call the deal.

If we discount this payment along with the total cash flows D2 holders received before the

call date by a discount rate of Treasury rate plus the pricing spread (71 bps), we have a

total present value of $14,427,071, or a price (priced to first call date) of $122.26!

Exhibit 124: Sample CLO Deal

Deal Information

Issue Date 3/29/01 WAC of Fixed Assets 8.85%

Reinvestment End Date 4/29/06 WAS of Floating Assets 2.92%

Non-Call End Date 4/29/06 Floating Rate Assets 74.33%

Legal Maturity 3/29/16 Payment Frequency Quarterly

Total Size 750,000,000

Capital Structure

Tranche Name Current Balance Spread/Coupon Rating Expiration Date of Make-Whole Premium

A1 563,500,000 L+47 bps Aaa N/A

A2 10,500,000 6.28% Aaa 4/29/2010

B1 40,000,000 L+125 bps A3 N/A

B2 22,500,000 6.95% A3 7/29/2011

C1 11,700,000 L+205 bps Baa2 N/A

C2 18,000,000 7.82% Baa2 10/29/2011

D1 12,000,000 L+635 bps Ba2 N/A

D2 11,800,000 12.05% Ba2 1/29/2012

Equity 60,000,000 N/A


Prepayment of HY Loans 5% CPR Prepayment of HY Bonds 20% CPR

Default of HY Loans 0.5% CDR Default of HY Bonds 2% CDR

Recovery of HY Loans 70% Recovery of HY Bonds 30%

Source: Credit Suisse, INTEX

We apply a similar calculation on other fixed tranches. For floating tranches, equity holders

typically only need to pay the par value of the aggregate outstanding amount to call. To

calculate the prices for floating tranches, we use the spreads consistent with new issue

pricing spreads. For comparison, we also calculate the prices assuming the deal is never

called. Exhibit 125 lists the results for all tranches. There are several interesting

observations:

1. For this deal to be called on the first redemption date (1 year later), the equity

holders will have to pay $655,246,986 (sum of Row 12) to the note holders.132

2. Equity holders will have to pay significant premiums on fixed tranches.

Therefore, the more fixed tranches in a deal, the less likely the deal will be

called, holding all else equal.

3. When priced to the first call date, the prices of all fixed tranches are higher

than the prices without call. It is the exact opposite for floating tranches.

132 We ignore any accrued interest, deferred interest , etc.

31 March 2006


Exhibit 125: Results of Base Case on the First Redemption Date (4th

Quarter from now)

Tranche Name

A1 A2 B1 B2 C1 C2 D1 D2

(1) WAL of Tranche if no call 3.05 3.05 6.11 6.11 6.93 6.93 7.69 7.69

(2) WAL of Tranche if called 0.91 0.91 0.91 0.91 0.91 0.91 0.91 0.91

(3)

Remaining Maturity to Make-

Whole Expiration Date (Year) 4.00 4.00 5.00 5.00 5.25 5.25 5.75 5.75

(4) Make-Whole Spread (bps) 77 111 155 371

(5)

Treasury Rate for Make-Whole

Premium (based on (3)) 3.82% 3.87% 3.87% 3.88%

(6)

Treasury Rate for Pricing if

Called (based on (2)) 3.40% 3.40% 3.40% 3.40%

(7)

Treasury Rate for Pricing if No

Call (based on (1)) 3.76% 3.88% 3.88% 3.89%

(8)

Pricing Spread over LIBOR (for

floating)/over Treasury (for fixed) 0.30% 0.71% 0.80% 1.22% 1.95% 2.25% 5.00% 5.20%

(9)

Coupon Rate (Fix)

/LIBOR Spread (Float) 0.47% 6.28% 1.25% 6.95% 2.05% 7.82% 6.35% 12.05%

(10) Make-Whole Premium 6.16% 8.68% 10.91% 20.68%

(11)

Remaining Notional

(on Redemption Date) 522,552,558 9,737,004 40,000,000 22,500,000 11,700,000 18,000,000 12,000,000 11,800,000

(12)

Optional Redemption Payout

= (11)*(1+(10)) 522,552,558 10,337,045 40,000,000 24,453,274 11,700,000 19,964,137 12,000,000 14,239,972

(13)

PV of Optional

Redemption Payout 501,367,043 9,922,886 38,186,354 23,355,416 11,041,699 18,874,896 10,986,126 13,078,431

(14) PV of cash flow before call 62,596,238 1,375,230 1,959,210 1,519,619 661,204 1,359,264 1,164,675 1,348,640

(15) Total PV of cash flow if called 563,963,281 11,298,116 40,145,564 24,875,035 11,702,903 20,234,161 12,150,801 14,427,071

(16) Price if called $100.08 $107.60 $100.36 $110.56 $100.02 $112.41 $101.26 $122.26

(17) Price if no call $100.28 $105.15 $102.15 $109.74 $100.31 $109.56 $107.21 $116.34


To call or not to call?

Which price then is the right price? Or, from a relative value perspective, which tranche,

floating or fixed, should investors buy and for how much? To answer these questions, we

assess the call likelihood of the deal.

Exhibit 126 shows the break-down of the numbers. By assuming a market price of $100,

the equity holders will collect $689,852,233 if they sell all the collateral. After paying the

termination fee of the swap contract and paying down all outstanding notes, the equity

holders are left with $33,733,390 on the redemption date, or a present value of

$29,923,236.133 Combined with the $11,614,423 present value of cash flows they received

from now to the redemption date, the total present value is $41,537,659, which provides

the left side of Equation 1.

133 Notice that in our example, there is still one year left from now to the first redemption date, or the end of

non-call period.

31 March 2006


Exhibit 126: Cash Flows on the First Redemption Date

Value

Asset Notional (on redemption date) $689,852,233

Market Price of Assets* $100.00

Market Value of Assets (on redemption date) $689,852,233

Swap Termination Payment (on redemption date) ($871,857)

Principal and Premium to Liabilities ($655,246,986)

Cash flow to Equity (on redemption date if called) $33,733,390

IRR of Equity** 12%

PV of cash flow to equity on call date $29,923,236

PV of cash flow to equity before call date $11,614,423

Total PV of cash flow to equity if called $41,537,659

PV of future fees (after call date) if not called $14,925,119

Total PV of cash flow to equity if not called $68,032,714

To Call or Not to Call? Not Call


* Assume dirty price with accrued interest for simplicity

** The current IRR available to equity holders from alternative investments

If the equity holders choose not to call the deal on the first redemption date, they will

receive future excess cash flows as well as management and incentive fees if they are

also the managers. By discounting the cash flows (including the ones they received from

now to the call date) and management and incentive fees at the equity IRR, they could

receive a total present value of $68,032,714, much higher than the expected proceeds if

they call the deal. Note the management and incentive fees may be a significant amount.

We assume the manager is not necessarily the majority equity holder and ignore the

management and incentive fees. In this example, the decision is not to call the deal

even without these fees. Therefore, from an economic perspective, this deal is unlikely to

be called on the first redemption date. If the deal is not called on the first call date but

is still priced to the first call date, the fixed tranches are over-valued while the

floating tranches are under-valued.

When’s the best time to call then?

Does this suggest tranches should be priced assuming no-call? No, because the deal may

be called later. Exhibit 127 shows whether to call or not to call on each payment

date/redemption date by number of quarters from today.

Exhibit 127: Call Decision at Each Redemption Date

Call Date (# of quarters from now) 4 8 12 13 14 15 16 20 24

PV Cash Flow to Equity If Called 41,537,659 46,857,301 51,100,208 51,942,908 52,426,590 53,099,680 53,505,206 54,401,451 54,305,781

PV of Cash Flow to Equity

If Not Called* 53,107,596 54,317,470 53,795,898 53,815,316 53,336,809 53,465,593 53,132,096 52,635,609 52,294,879

Call or Not Call Not Call Not Call Not Call Not Call Not Call Not Call Call Call Call


* Ignore management and incentive fees

It shows, keeping everything else constant, the present value of call proceeds to equity

holders will be higher than the present value of cash flows if the deal is called 16 quarters

from now. Only on this redemption date, for the first time, will it be economically optimal to

call the deal.134

134 If it is not called on the 16th quarter from now, it may still be optimal to call on the later redemption

dates.

31 March 2006


What’s the fair value for the tranches?

Exhibit 128 shows the detailed information on the 16th quarter (from today), when the deal

should be called. The prices are very similar for A1 and A2 tranches whether called or not,

as the majority of the balance of both tranches have been paid down by then.135 The fair

price for these two tranches should be $100.25 and $105.53 respectively, for a DM of 30

bps for the A1 and a spread pick-up over Treasury of 71 bps for the A2, both with a WAL

of 2.75 years.

We examine the prices against the redemption dates to see how the prices change at

each date and find the upper and lower bounds of the fair price. We use Tranche D2 as an

example. Exhibit 129 lists the price of D2 called on each redemption date, as well as the

price without call. After a certain period, prices converge as the tranche pays down before

the potential redemption date. Whether the deal is called or not, the last row lists the final

fair price for the tranche. Exhibit 130 plots the prices against time (in terms of number of

quarters from today). There are a couple of interesting findings:

1. For Tranche D2, the price to call can actually drop below the price without the

call.136 The lower boundary is around $115 at the 27th quarter, when the make-

whole premium expires and the equity holders only have to pay par to call.

2. The upper boundary for “fair” price should be about $118.25 when it should be

called for the first time 16 quarters from today.

Without make-whole premium, more likely to call

One of the advantages our model provides is flexible scenario analysis. The first scenario

removes the make-whole premium. As shown in Exhibit 131, without the make-whole

premium, the deal could be called much earlier – in the 12th quarter, rather than the 16th.

How Much Tightening in CDO Liability Spreads To Trigger a Call?

We believe a large part of the appreciation in the underlying collateral is being driven by

CDO buying.137 As the increasing demand for CLO paper is reflected in the tightening of

CLO spreads, we believe there is a connection between the appreciation of leverage loan

prices, or tightening in leverage loan spreads, and the tightening in CLO liability spreads.

A simple regression of CLO liability spreads on leverage loan spreads shows the

correlation is 63%.138 More interestingly, the regression also shows that for each basis

point of tightening in CLO liability spreads,139

the leverage loan spread tightens by

0.89 basis point.

To answer this question, we use our sample deal as an example. We first stress the CLO

liability spread tightening. Then we calculate the spread tightening of leverage loans based

on the regression result aforementioned. As leverage loans are floating rate and based on

our calculation, we use a spread dollar duration of $0.1140 for each basis point change in

spread. Finally we calculate the prices under each scenario and see how much tightening

in CLO liability spread will trigger a call.

135 As a matter of fact, if the entire tranche is paid off before the call date, the prices will be exactly the

same. 136

All the prices calculated are impacted by the shape of the Treasury curve, as the Treasury rates used are all WAL adjusted. 137

Based on CSFB's estimate, more than 60% of the institutional loans are gobbled by CLOs. 138

Data sample used include monthly numbers from late 2001 to now. The leverage loans spread used is the CSFB Leverage Loan Index. 139

Weighted average spread of CLOs. 140

In other words, for each basis point change in spread, the price of the leverage loan will change by $0.1.

31 March 2006


Exhibit 128: Results of Base Case at the 16th

Quarter from Today

Tranche Name

A1 A2 B1 B2 C1 C2 D1 D2

(1) WAL of Tranche if no call 3.05 3.05 6.11 6.11 6.93 6.93 7.69 7.69

(2) WAL of Tranche if called 2.75 2.75 3.91 3.91 3.91 3.91 3.91 3.91

(3)

Remaining Maturity to Make-Whole

Expiration Date (Year) 1 1 2 2 2.25 2.25 2.75 2.75

(4) Make-Whole Spread (bps) 77 111 155 371

(5)

Treasury Rate for Make-Whole

Premium (based on (3)) 3.42% 3.66% 3.66% 3.66%

(6)

Treasury Rate for Pricing if Called

(based on (2)) 3.66% 3.80% 3.80% 3.80%

(7)

Treasury Rate for Pricing if No Call

(based on (1)) 3.76% 3.88% 3.88% 3.89%

(8)

Pricing Spread over LIBOR (for

floating)/over Treasury (for fixed) 0.30% 0.71% 0.80% 1.22% 1.95% 2.25% 5.00% 5.20%

(9)

Coupon Rate (Fix)

/LIBOR Spread (Float) 0.47% 6.28% 1.25% 6.95% 2.05% 7.82% 6.35% 12.05%

(10) Make-Whole Premium 2.04% 4.14% 5.51% 11.56%

(11)

Remaining Notional

(on Redemption Date) 160,402,257 2,988,862 40,000,000 22,500,000 11,700,000 18,000,000 12,000,000 11,800,000

(12)

Optional Redemption Payout

= (11)*(1+(10)) 160,402,257 3,049,727 40,000,000 23,431,257 11,700,000 18,992,157 12,000,000 13,163,979

(13)

PV of Optional

Redemption Payout 133,401,390 2,563,258 32,606,465 19,192,517 9,108,497 14,936,949 8,274,100 9,220,917

(14) PV of cash flow before call 431,527,312 8,517,200 7,963,455 5,635,141 2,607,788 4,967,784 4,245,175 4,732,308

(15) Total PV of cash flow if called 564,928,702 11,080,458 40,569,920 24,827,658 11,716,285 19,904,733 12,519,275 13,953,225

(16) Price if called $100.25 $105.53 $101.42 $110.35 $100.14 $110.58 $104.33 $118.25

(17) Price if no call $100.28 $105.15 $102.15 $109.74 $100.31 $109.56 $107.21 $116.34


Exhibit 129: Tranche D2 Valuation

Call Date (# of quarters from now) 4 8 12 14 15 16 18 20 24 27 28 32 36

Call or Not Call Not Call Not Call Not Call Not Call Not Call Call Call Call Call Call Call Call Not Call

Price of D2 if called 122.26 121.36 119.65 119.16 118.88 118.25 117.63 117.25 115.85 114.92 115.27 116.30 116.34

Price of D2 if not called ever 116.34 116.34 116.34 116.34 116.34 116.34 116.34 116.34 116.34 116.34 116.34 116.34 116.34

Price adjusted by Call 116.34 116.34 116.34 116.34 116.34 118.25 117.63 117.25 115.85 114.92 115.27 116.34 116.34

Source: Credit Suisse , INTEX

31 March 2006


Exhibit 130: Price of Tranche D2

114

115

116

117

118

119

120

121

122

123

0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38

# of Quarters from Now

Pri

ce

($

)

Price of D2 if called Price of D2 if not called ever Fair Price adjusted by Call


Exhibit 131: Call Decision on Each Redemption Date (without Make-Whole Premium)

Call Date (# of quarters from now) 4 8 12 16 20 24 28

PV Cash Flow to Equity If Called 47,709,249 51,407,980 54,275,824 55,577,602 55,551,519 54,590,744 53,411,659

PV of Cash Flow to Equity If Not Called 53,107,596 54,317,470 53,795,898 53,132,096 52,635,609 52,294,879 52,162,267

Call or Not Call Not Call Not Call Call Call Call Call Call


Exhibit 132 shows the results on the first redemption date four quarters from now.

Although at a market price of $100 it is not economically rational to call the deal, it is the

case if CLO liability spread tightens by 22 bps, causing the market price to jump to

$101.78. This change will also impact the final fair price of the tranches because: 1) it

changes the probability of call, which, in turn, determines which price to use − the one

priced to call or the one sans call; and 2) it changes the discount rate used, as we use the

new issue spread for pricing.

We can also run the same analysis on a different potential redemption date and create a

two-dimensional (liability spread tightening vs. time) matrix.

31 March 2006


Exhibit 132: Call Decision on First Redemption Date by Changing CLO Liability Spreads CLO Liability Spread

Tightening (bps) -10 -5 5 10 21 22 23 25 30

Leverage Loan Spread

Tightening (bps) -8.90 -4.45 - 4.45 8.90 17.80 18.69 19.58 20.47

Market Price $99.11 $99.56 $100.00 $100.45 $100.89 $101.78 $101.87 $101.96 $102.05

PV Cash Flow to Equity

If Called $36,091,446 $38,845,149 $ 44,291,362 $ 46,983,872 $52,980,826 $53,531,567 $54,082,308 $55,183,789 $56,224,077

PV of Cash Flow to Equity

If Not Called $53,107,596 $53,107,596 $ 53,107,596 $ 53,107,596 $53,107,596 $53,107,596 $53,107,596 $53,107,596 $53,107,596

Call or Not Call Not Call Not Call Not Call Not Call Not Call Call Call Call Call


Higher IRR on Alternative Investments for Equity Holders, More Likely to Call

If equity holders can earn a higher IRR on alternative investments, a higher discount rate

is used to calculate the present value of future cash flows to the equity holders if not called,

or the right side of Equation 1. The lower the present value, the more likely equity holders

call the deal. In other words, if the equity holders have limited investment options, the deal

is less likely to be called.

Faster Prepayment Speed, More Likely to Call

So far we used base line prepayment and default assumptions: 5 CPR and 2 CDR for

bonds and 20 CPR and 0.5 CDR for loans. What if the expected prepayment speed is

faster? Given a sequential payment schedule, faster prepayments may reduce the

leverage and increase the funding cost faster. Also, faster prepayments retire the notes

earlier, reducing potential management and incentive fees collected in the future. All these

factors contribute to a higher likelihood of calling the deal when prepayment jumps.

Closing Thoughts Assessing the call risk of CDOs is no easy task, even from a pure economic perspective.

The diversity of CDO structures and the unpredictability of market conditions make it

difficult to precisely estimate the probability of call and evaluate a CDO tranche’s fair value

from a macro perspective. We take a micro approach by leveraging the CDO modeling

expertise and CDO deals covered by Intex, running projected cash flows through our CDO

call model. Based on this model, we can gauge the impact and sensitivity of each

parameter on CDO calls and reasonably estimate the fair value of a certain tranche. The

market might over- or under-estimate the call probability from time to time. If a floating

tranche is priced to call but the call probability is very low, the tranche will be undervalued;

if a fixed tranche is priced to call but the model shows the call is not very likely, the fixed

tranche is usually overvalued, especially if it has a more generous make-whole premium.

In addition, typically the room for profit is bigger for lower-rated tranches.

31 March 2006


A Comparison of US and European CLOs141

The European CLO market has seen tremendous growth in recent years, echoing the

growth of the US CLO market.142 In this section, we take a comprehensive look at these

two markets, comparing the underlying collateral, deal-level characteristics and market

conditions. The objectives of this piece are to provide investors with a greater

understanding of both the US and European CLO markets, to keep readers abreast of new

developments in the CLO markets, and to advocate continued improvement of analytical

and valuation capabilities.

The collateral: US vs. European leveraged loan markets First let’s look at the underlying collateral, the institutional leveraged loan markets.

The issuance of both US and European leveraged loans have increased dramatically

since 2003, yet driven by different factors. In Europe, LBOs (Leveraged Buy Outs) have

been the primary force behind the surge in leveraged loan issuance, while in the US, other

factors, such as M&A activity and refinancing due to the low interest rate environment,

have contributed to the substantial growth.

Exhibit 133: New Issue Institutional Leveraged Loan Volume: US vs. Europe

55 50

32

64

118

223

152

5 7 11 13 1326

55

0

50

100

150

200

250

1999 2000 2001 2002 2003 2004 Jan-Oct

2005New

In

sti

tuti

on

al L

oan

Vo

lum

e (

$ B

illio

ns

)

US ($BN) Europe ($BN)

Source: CREDIT SUISSE, LPC, S&P LCD. European figures converted to USD using Euro spot rate of 1.1953.

Exhibit 133 shows the total volume of newly issued institutional leveraged loans in the US

and Europe. Compared to the US market, the European market is still relatively small: $26

billion versus $223 billion in 2004. However, in terms of the pace of growth, the European

market is equally as impressive.143

Unlike its US counterpart, most of the proceeds from European leveraged loan issuance

are used for LBO financing, as shown in Exhibit 134 and Exhibit 135. An astonishing 95%

of European issuance during the first half of 2005 was driven by LBOs. This phenomena

has critical implications for analyzing the credit risk of European loans as LBO-motivated

issuances tend to have more aggressive credit characteristics.

141 This section was originally published in "The CDO Strategist", Issue #12, December 15, 2005.

142 In this piece, we use "CLO" and "CDO" interchangeably.

143 For a further discussion of European and US leveraged loan markets, please refer to CSFB’s Global

Leveraged Finance Strategy & Portfolio Products report: The Week in Leveraged Finance, Week Ending

June 9, 2005, by Sam DeRosa-Farag and team.

Leveraged loan

issuance strong in

both the US &

Europe

LBO drives

European loan

market

31 March 2006


The higher risk embedded in European loans can be exhibited by comparing the leverage

ratios. Both senior debt multiple and total debt multiple are higher for European loans than

for US loans (see Exhibit 134 and Exhibit 135). Especially noteworthy is the debt multiple of

European loans jumping significantly, to 5.5 times EBITDA in recent months, a 7-year high.

Exhibit 136: Comparing Leverage: Average Senior Debt Multiples

3.6

4.3

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

4.5

5.0

Mar-98

Jul-98

Nov-98

Mar-99

Jul-99

Nov-99

Mar-00

Jul-00

Nov-00

Mar-01

Jul-01

Nov-01

Mar-02

Jul-02

Nov-02

Mar-03

Jul-03

Nov-03

Mar-04

Jul-04

Nov-04

Mar-05

Avera

ge S

en

ior

Deb

t/E

BIT

DA

US Europe

Source: CREDIT SUISSE, S&P LCD

Exhibit 137: Comparing Leverage: Average Total Debt Multiples

4.0

5.5

3.0

3.5

4.0

4.5

5.0

5.5

6.0

Mar-98

Jul-98

Nov-98

Mar-99

Jul-99

Nov-99

Mar-00

Jul-00

Nov-00

Mar-01

Jul-01

Nov-01

Mar-02

Jul-02

Nov-02

Mar-03

Jul-03

Nov-03

Mar-04

Jul-04

Nov-04

Mar-05

Avera

ge T

ota

l D

eb

t/E

BIT

DA

US Europe


Exhibit 134: US Leveraged Loans: Use of Proceeds*

Exhibit 135: Euro Leveraged Loans: Use of Proceeds*

M &A

20%

Recap/

Dividend

20%

Refinancing

33%Recap/ Other

14%

Recap/ IPO

5%

Other

8%

Refinancing

5%

LBO

95%


* During 1st

half of 2005


* During 1st

half of 2005

Higher leverage for

European loans

31 March 2006


There is a remarkable difference between the US and European leveraged loan markets in

terms of loan pricing and spread performance. As shown in Exhibit 138, the pricing and

spread movement for European loans has been very static, whereas US loan spreads

have tightened dramatically since mid-2003.

In addition, given BB spreads on top of B spreads for European loans suggests European

loan pricing is driven by factors other than credit ratings. That said, we note a recent growing

differential in pricing, with BB spreads about 20 bps (272 vs. 292) tighter than B spread.

Exhibit 138: Institutional Loan New Issue Spreads

269

292

187

272

180

230

280

330

380

430

Dec-99

Mar-00

Jun-00

Sep-00

Dec-00

Mar-01

Jun-01

Sep-01

Dec-01

Mar-02

Jun-02

Sep-02

Dec-02

Mar-03

Jun-03

Sep-03

Dec-03

Mar-04

Jun-04

Sep-04

Dec-04

Mar-05

Jun-05

New

Issu

e S

pre

ad

s (

ove

r L

IBO

R)

US B Spread Euro B Spread US BB Spread Euro BB Spread

Euro BB wider than B

Back to "normal":

B wider than BB


With this, CDOs benefit from illiquidity and pricing discrepancies; the arbitrage between

asset and liability for European CLOs is higher than for US CLOs, improving CDO

economics in Europe. This attractive arbitrage exists because while US and European

CLO liability costs have dropped in tandem, European leveraged loan spreads have

remained wider – a point we revisit later. Note that the spread difference between

European BB-rated loans and US BB-rated loans is 85 bps, a very attractive pick-up

assuming all things equal.

Unlike the US market, the European loan market remains largely private in nature as a

significant share of loans are unrated and held by traditional banks. Exhibit 139 and

Exhibit 140 compares the rating distributions of US and European loans based on the

CREDIT SUISSE US and Western European Institutional Leveraged Loan Index; nearly

47% of European loans are unrated.

European loan

spreads more static

than US loan

spreads; pricing

abnormalities exist

More European

loans are unrated

Exhibit 139: US Institutional Loans by Rating* Exhibit 140: European Institutional Loans by Rating*

B

39%

Split BB

10%Split BBB

6%CCC/Split

CCC

8%

NR

1%

Split B

9%

BB

27%

CCC/Split

CCC

0.1%

NR

46.8%

Split B

1.0%

Split BB

11.8%

BB

8.7%

B

30.9%

Split BBB

0.7%

Source: Credit Suisse, As of 11/30/2005.

* Based on CREDIT SUISSE US Institutional Loan Index


* Based on CREDIT SUISSE European Institutional Loan Index

31 March 2006


It is helpful to understand who the participants are in leveraged loans of both jurisdictions.

Exhibit 141 and Exhibit 142 show the investor bases of all leveraged loans (institutional

and pro-rata tranches). 144 Clearly, the biggest investors of US leveraged loans are CDOs,

while European banks dominate the European loan market with a 62% share. This is due

to the composition of leveraged loans in both markets: the US market has a much larger

institutional share of leveraged loans while the European market remains predominantly

pro rata (see Exhibit 143 and Exhibit 144). This helps explains why European institutional

loan pricing has been by and large static; most of the market remains in the private sector,

minimizing price fluctuations and credit rating differentiation. However, we note that the

institutional share in the European loan market has continued to grow, likely due to the

proliferation of institutional investors and CDOs in the market.

Because CLOs invest mostly in institutional loans, their market share looks much more

similar between the jurisdictions when only considering the institutional loan portion: CLOs

dominate the US institutional loan market with 61% and also dominate the European

market with a more impressive 83% (as of 1H 2005).145

144 Leveraged loans are typically structured with a pro rata portion, comprising a revolving facility and a

Term Loan A (TLA), and an institutional loan portion, comprising Term Loan B (TLB), Term Loan C (TLC) or other tranches. 145

CSFB Global Leveraged Finance Strategy & Portfolio Products, and S&P LCD.

CDOs dominate US

leveraged loan

market while banks

dominate European

leveraged loan

market

CLOs dominate the

institutional loan

markets

Exhibit 141: Investors of ALL US Leveraged Loans Exhibit 142: Investors of ALL European Lev. Loans

European

Banks

8%

Insurance Co.

4%

Other

5%

Finance Co.

6%

Domestic

Banks

10%

CDOs &

Hedge/HY

Funds

67%

CDOs

19%

US Banks

5%

European Banks

62%

Other

12% Insurance Co.

1%

Finance Co.

1%

Source: Credit Suisse, S&P LCD, As of 1H 2005. Source: Credit Suisse, S&P LCD, As of 1H 2005.

Exhibit 143: US Pro-Rata vs. Institutional ‘01-‘05 Exhibit 144: Europe Pro-Rata vs. Institutional ‘01-‘05

75%

60%

44% 40% 38%

25%

40%

56% 60% 62%

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

2001 2002 2003 2004 2005*

Pro-Rata Institutional

89%

73% 77%66%

55%

11%

27% 23%34%

45%

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

2001 2002 2003 2004 2005*

Pro-Rata Institutional

Source: Credit Suisse, S&P LCD. * As of Oct 2005. Source: Credit Suisse, S&P LCD. * As of Oct 2005.

31 March 2006


It is also interesting to see the industry breakdown of both loan markets, as shown in

Exhibit 145 and Exhibit 146. “Media and Telecom” stands out as the biggest bucket in both

jurisdictions: 19% in US and 36% in Europe.

How have both US and European loans performed so far? Unfortunately, the performance

data on European loans is very limited, however, we can gain some perspective by

observing the ratings transition matrix. Exhibit 147 and Exhibit 148 show the average one-

year rating transitions of both US and European loans. Although the comparison isn’t

exactly “apples-to-apples” (i.e. the US matrix is based on a much longer history – 20 years

from 1984 to 2004 – while the European numbers are only based on the performance

during 2003 and 2004) we can glean some general ideas.

Media and Telecom:

Lion’s Share

Exhibit 145: US Institutional Loan by Industry* Exhibit 146: European Institutional Loan by Industry*

Healthcare

7%

IT

5%

Transportat ion

9%

Service

3%

Retail

2%

M etals/M ineral

4%

Chemicals

5%

Utility

10%

Food & Drug

2%

Food/tobacco

3%

Forest

Prod/Container

6%

Gaming/Leisure

6%

Housing

3%M fcturing

3%

M edia/Telecom

19%

Consumer

Products

3%

Energy

7%

Financial

2%

Aerospace

2%

Food/tobacco

4.0%

Chemicals

9.6%

Aerospace

1.1%

M edia/Telecom

35.9%

M etals/M ineral

0.1%

Retail

6.8%

M fcturing

6.4%

Housing

4.7%Healthcare

3.2%

Gaming/Leisure

3.0%

Service

9.1%

Transportat ion

4.1%

Utility

0.5% Consumer

Durables

1.4%

Consumer Non-

durables

2.1%

Forest

Prod/Container

6.1%

Financial

1.7%

Energy

0.3%


* Based on CREDIT SUISSE US Institutional Loan Index


* Based on CREDIT SUISSE European Institutional Loan Index

Exhibit 147: Average 1-Year Transition Rates of US Loans, 1984-2004 (%)

From/To BB+ BB BB- B+ B B- CCC D Downgrade

Upgrade or

Stable Ratio

BB+ 69.6 6.8 3.7 1.6 0.9 0.2 0.7 0.6 14.5 69.6 4.8

BB 7.5 71.8 8.2 3.2 1.8 0.5 1.0 1.0 15.7 79.3 5.1

BB- 2.5 8.0 71.4 9.0 3.2 1.2 1.3 1.9 16.6 81.9 4.9

B+ 0.4 1.7 6.0 75.8 6.8 2.7 2.6 3.4 15.5 83.9 5.4

B 0.5 0.7 2.0 8.6 65.6 6.4 6.2 9.5 22.1 77.4 3.5

B- 0.3 0.2 0.8 4.1 8.1 60.1 11.7 14.0 25.7 73.6 2.9

CCC/C 0.2 0.4 1.0 1.3 3.1 6.7 53.5 33.0 33.0 66.2 2.0

Source: S&P

Exhibit 148: Average 1-Year Transition Rates of European Loans, 2003-2004 (%)

From/To BBB BBB- BB+ BB BB- B+ B B- CCC D Downgrade

Upgrade

or Stable Ratio

BB+ 5.6 0.0 52.8 23.6 11.8 6.3 0.0 0.0 0.0 0.0 41.7 58.4 1.4

BB 0.0 0.0 2.7 73.8 12.0 5.8 5.8 0.0 0.0 0.0 23.6 76.5 3.2

BB- 0.0 0.0 0.0 1.6 82.3 11.3 1.6 3.2 0.0 0.0 16.1 83.9 5.2

B+ 0.0 0.0 0.7 0.0 5.4 77.7 9.5 4.1 2.7 0.0 16.3 83.8 5.1

B 1.9 0.0 0.0 0.0 0.0 6.8 71.6 13.5 6.4 0.0 19.9 80.3 4.0

B- 0.0 0.0 1.9 0.0 0.0 1.9 9.4 71.8 8.3 6.6 14.9 85.0 5.7

CCC 0.0 0.0 0.0 0.0 0.0 0.0 0.0 13.9 66.2 20.0 20.0 80.1 4.0

Source: S&P

31 March 2006


For example, “Upgrade/Stable versus Downgrade” ratios suggest that at BB+/BB levels,

US loans outperform European loans, while at B and below levels, Europe outperforms

US.146 This observation seems consistent with the spreads shown in Exhibit 139 B-rated

European loans are priced to similar levels as BB-rated loans, perhaps because their

credit quality is similar to BB-rated loans. 147 We note that this is just one possible

explanation; with more empirical evidence, stronger conclusions may be drawn.

Deal Level: US versus European CLOs Both US and European CLO markets have exhibited significant growth and expansion in the

past several years. This has been driven mainly by increasing investor demand for CLO

paper due to superior credit performance. In 2005 so far, approximately $42 billion of US

CLOs have been issued, up 68% versus 2004 volumes; the European CLO market, while

small compared to the US, has seen an even more impressive growth rate, up 93% from

$6.9 billion last year to $13.3 billion in 2005 to date (see Exhibit 149).

Exhibit 149: CLO Volume: US vs. Europe

19.117.1

13.5 14.4

17.7

24.9

41.8

0.42.0

3.5 3.4 4.56.9

13.3

0.0

5.0

10.0

15.0

20.0

25.0

30.0

35.0

40.0

45.0

1999 2000 2001 2002 2003 2004 Up thru

11/2005

CL

O V

olu

me (

$ B

illio

ns)

US CLO ($BN) Europe CLO ($BN)


Exhibit 150 and Exhibit 151 show the top 15 CLO managers for both jurisdictions. With

respect to these managers, there is minimal overlap, with the exception of managers such

as Babson Capital, Invesco and PIMCO. We have also seen some new managers entering

the European CLO market, such as The Carlyle Group, Rabobank, CELF Investment

Advisors, GSC Partners, WestLB, AIB Capital Markets, and CSAM.

146 The higher the ratio, the more loans stay stable or get upgraded, than loans get downgraded.

147 The order of the ratio of US loans seems to be reasonable: higher rated loans have better performance,

i.e., higher ratios; while the order of European loans seems to be counterintuitive.

Rating transition

performance mixed

Robust CLO

issuances

Top CLO managers

31 March 2006


Exhibit 150: Top 15 CLO Managers in US (all vintages)

Rank US CLO Manager Total Issuance ($BN) Deal Count

1 Highland Capital Management 8.63 12

2 Babson Capital 8.19 15

3 Credit Suisse Asset Management 7.57 13

4 ING Capital Advisors 5.33 10

5 Invesco Institutional Inc. 4.85 10

6 Stanfield Partners LLC 4.84 9

7 Ares Management 4.65 10

8 Sankaty Advisors 4.52 10

9 American Express Asset Management Group 4.35 7

10 Black Diamond Capital Management 4.30 6

11 TCW Asset Management 3.63 8

12 Pacific Investment Management Company 3.41 8

13 Deerfield Capital Management 2.92 8

14 Chase Capital Partners (Octagon Credit Investors) 2.79 6

15 ING Pilgrim 2.78 6


Exhibit 151: Top 15 CLO Managers in Europe (all vintages)

Rank European CLO Manager Total Issuance ($BN) Deal Count

1 Harbourmaster Capital 3.03 5

2 Alcentra Group 2.62 5

3 Babson Capital Europe Limited 2.57 5

4 Intermediate Capital Group 2.43 8

5 Pacific Investment Management Company 2.06 3

6 AXA Investment Managers 1.60 4

7 Avoca Capital 1.36 3

8 Mizuho Corporate Finance 1.35 2

9 Allied Irish Bank Capital Markets 1.27 3

10 Prudential M&G 1.13 3

11 CELF Investment Advisors 1.13 2

12 Invesco 1.11 3

13 RMF Investment Products 1.11 3

14 NIB Capital Management 0.93 2

15 BNP Paribas 0.89 3


CLO spreads have been tightening in both markets. Senior spreads of European CLOs have

converged with US spreads while subordinate spreads at the BBB level have compressed

even tighter than US levels. Exhibit 152 and Exhibit 153 show historical CLO spreads for

AAA and BBB tranches.

CLO spreads

tightening in both

markets

31 March 2006


Exhibit 152: AAA CLO Spreads: US versus Europe*

15

25

35

45

55

65

75

Sep-01

Dec-01

Mar-02

Jun-02

Sep-02

Dec-02

Mar-03

Jun-03

Sep-03

Dec-03

Mar-04

Jun-04

Sep-04

Dec-04

Mar-05

Jun-05

Sep-05

Dec-05

AA

A C

LO

Sp

rea

ds

(b

ps

)

US CLO AAA

EU CLO AAA

Source: CREDIT SUISSE, Intex

* US spread over LIBOR, Europe spreads over Euribor

Exhibit 153: BBB CLO Spreads: US versus Europe*

90

140

190

240

290

340

Sep-01

Dec-01

Mar-02

Jun-02

Sep-02

Dec-02

Mar-03

Jun-03

Sep-03

Dec-03

Mar-04

Jun-04

Sep-04

Dec-04

Mar-05

Jun-05

Sep-05

Dec-05

BB

B C

LO

Sp

read

s (

bp

s)

US CLO BBB

EU CLO BBB



There has been a convergence in CLO all-in liability costs too. As shown in Exhibit 154, the

all-in liability cost of European CLOs is similar to that of US CLOs: both at around 36 bps. As

noted earlier, with underlying loan spreads for European loans wider than US loan spreads,

the arbitrage is higher for European CLOs. Assuming all things equal, this means higher

potential IRR for European CLO equity investors.

Exhibit 154: CLO All-in Liability Cost*

0

10

20

30

40

50

60

70

80

90

2H01 1H02 2H02 1H03 2H03 1H04 2H04 1H05 6ME Oct

05

CL

O C

os

t o

f F

un

din

g (

bp

s)

EU CLO Cost of Funds

US CLO Cost of Funds

Assuming the follow ing capital structure: 70% AAA, 10% AA, 3% A & 7% BBB

Source: CREDIT SUISSE, Intex, S&P


31 March 2006


There has been a growing concern regarding collateral over-concentration in CLOs. To

address this concern, we compare the collateral industry breakdown and rating breakdown

of US CLOs versus European CLOs.

Exhibit 155: US CLO Collateral by Ratings* Exhibit 156: European CLO Collateral by Ratings*

A

0.1%

BB

35.6%

B

56.1%

UNRATED

4.3%CCC & Below

2.2%

BBB

1.7%

BBB

0.4%

B

38.8%

CCC & Below

0.2%

UNRATED

38.3%

BB

22.2%

Source: Credit Suisse, Intex.

* 2004 – 2005 vintages.


* 2004 – 2005 vintages

We also looked at rating distribution. As shown in Exhibit 155 and Exhibit 156, the most

noticeable difference there are a lot more unrated loans in Europe, which poses additional

challenge when analyzing European CLOs.

Exhibit 157 and Exhibit 158 show the industry distribution of collateral in 2004 and 2005

vintage US and European CLOs available in Intex. Two general conclusions can be drawn

from these charts:

1. The industry distributions are different between US and European CLOs. The top

three industries in US CLOs are 1) Healthcare, Education & Childcare; 2)

Broadcasting & Entertainment; and 3) Conglomerate Manufacturers. However, the

top 3 industries in European CLOs are 1) Transportation; 2) Telecom; and 3)

Chemicals.

2. On an aggregate level, there is no industry over-concentration as the collateral is

highly diversified across all industries: no single industry has a share exceeding

10%.

More unrated loans

in European CLOs

Exhibit 157: US CLO Collateral by Industries* Exhibit 158: European CLO Collateral by Industries*

Broadcast ing &

Ent ert ainment

7%

Leisure

5%

Lodging

3%

Nat ural Resources

3%

Text i les

1%

Transport at ion

5%

Telecom

5%Ret ai l

3%

Building & Real Est at e

6%

Cont ainers

4%

Consumer Non-

Durables

3%

Agricult ure

1%

Aerospace & Def ense

3%Ut i l i t ies

4%

Ot her

15%

Machinery

1%

Oil & Gas

4%

Conglomerat e Mfg.

6%

Chemicals

5%

Healt hcare, Educat ion

& Childcare

8%

Grocery

0%

Food

6%

Telecom

9%

Tex t iles

1% Conglomerat e Mfg.

4%

Healt hcare, Educat ion &

Chi ldcare

4%Indust r ial

0%

Leisure

5%

Consumer Durables

0%

Chemicals

9%

Ut il i t ies

3%

Transpor t at ion

10%

Machinery

1% Lodging

3%

Nat ural Resources

1%

Oil & Gas

2%

Ret ail

4%

Ot her

17%

Aerospace & Defense

1%

Grocery

1%

Agr icult ure

1%Business Eq. & Serv ices

0%

Consumer Non-Durables

3%Cont ainers

6%Food

6%

Broadcast ing &

Ent ert ainment

5%

Building & Real Est at e

5%

Source: Credit Suisse, Intex.

* 2004 – 2005 vintages.


* 2004 – 2005 vintages

No industry over-

concentration found

31 March 2006


To address the overlapping of issuers among CLO pools, we reviewed each asset in all

CLOs issued from 2004 to 2005 of which collateral information was obtainable, and counted

the number of CLOs sharing identical loan issuers in their portfolios. Our sample includes 73

US CLOs and 16 European CLOs and Exhibit 159 and Exhibit 160 show the lists of issuers

appearing in more than half of the CLOs.

Exhibit 159: Issuer Concentration Among US CLOs (2004-2005 Vintage)

Rank Issuer Name

Number of CLOs

with This Issuer Percent* RankIssuer Name

Number of CLOs

with This Issuer Percent*

1 MGM 67 92% 22 Direct TV 45 62%

2 Kerr Mcgee 64 88% 23 Movie Gallery 45 62%

3 General Growth Properties 60 82% 24 Smurfit Stone Container 45 62%

4 Graham Packaging 56 77% 25 Dresser Rand Group 44 60%

5 Boise Cascade 55 75% 26 Novelis 44 60%

6 Jean Coutu Group 54 74% 27 Venetian Casino 44 60%

7 Panamsat Corp 54 74% 28 BCP Caylux 43 59%

8 Huntsman Corp 51 70% 29 Lake Las Vegas Resort 43 59%

9 Constellation 50 68% 30 MCC Iowa 42 58%

10 Regal Cinemas Inc 49 67% 31 Invensys International 41 56%

11 Reliant Energy 49 67% 32 Universal City Development Partners 41 56%

12 Resort International 49 67% 33 Charter Communications 39 53%

13 Rockwood Specialties Group 49 67% 34 Pinnacle Foods 39 53%

14 Goodyear Tire & Rubber 48 66% 35 Community Health 38 52%

15 Texas Genco 48 66% 36 Foundation Coal 38 52%

16 Valor Telecommunications 48 66% 37 Jarden 38 52%

17 Fidelity National Information Solutions 47 64% 38 Nortek 38 52%

18 UGS 47 64% 39 Spectrum Brands 38 52%

19 Allied Waste 46 63% 40 Cooper Standard 37 51%

20 R.H. Donnelley 46 63% 41 Hercules Offshore 37 51%

21 Warner Chilcott Holdings Company 46 63%


* Divided by 73, the total number of US CLOs in our sample

It is somewhat disconcerting to see some names appearing in most of the deals, for example,

MGM is in 67 out of 73 US CLOs and TDF appears in 13 out of 16 European CLOs.

However, on an aggregate basis, the overlap does not seems as significant as anticipated.

Note that there are about 1,300 issuers in all sampled US CLOs and about 480 issuers in all

sampled European CLOs. But only 41 and 25 issuers, US and European respectively,

appear in more than half of the sampled CLOs.

Another challenge facing investors in European CLOs is the fact that the insolvency regimes

and bankruptcy procedures among European countries could be significantly different. For

example, the laws are relatively more creditor-friendly in the UK while they tend to be less so

in France. These differences have important implications on issues such as the recovery

rate of leveraged loans and CLOs. For example, S&P assumes different recovery rates for

different European jurisdictions, as show Exhibit 161.

Issuer overlap not as

significant as

expected

Insolvency regimes

differ among

countries

31 March 2006


Exhibit 160: Issuer Concentration of European CLOs (2004-2005 Vintage)

Rank Issuer Name

Number of CLOs

with This Issuer Percent Rank Issuer Name

Number of CLOs

with This Issuer Percent

1 TDF 13 81% 14 Fina Cold 9 56%

2 Cognis Deutschland II 12 75% 15 Ineos Acrylics Finance 9 56%

3 Grohe Holdings 11 69% 16 Rockwood Specialties 9 56%

4 Kabel Deutschland 11 69% 17 WAM Acquisition 9 56%

5 Satbirds Finance 11 69% 18 World Directories Acquisition 9 56%

6 Aster 10 63% 19 Corleone Capital 8 50%

7 Frans Bonhomme 10 63% 20 Elis Group 8 50%

8 Kappa Packaging 10 63% 21 Invensys International Holdings 8 50%

9 Tank & Rast 10 63% 22 Materis Holding Luxembourg 8 50%

10 Debitel 9 56% 23 Nachtwache Acq 8 50%

11 Demag Investments 9 56% 24 OGF Holding 8 50%

12 Dragoco Gerberding 9 56% 25 Springer Science & Business Media 8 50%

13 Editis 9 56%


* Divided by 16, the total number of European CLOs in our sample

Exhibit 161: S&P Recovery Rate Matrix by Country

S&P Priority Category US Group 1 Group 2 Group 3 Group 4

Senior Secured Loans 56% 60-75% 50-60% 45-55% 40-50%

Senior Unsecured Loans 40% 35-45% 35-45% 30-40% 25-35%

Subordinated Loans 22.8% 10-15% 20-30% 15-25% 10-20%

S&P Priority Category US Group 1 Group 2 Group 3 Group 4

Group 1: Ireland & UK

Group 2: Germany, Netherlands & Switzerland

Group 3: Austria, Denmark, Finland, Norway & Sweden

Group 4: Belgium, France, Greece, Italy, Luxembourg, Portugal & Spain

Source: S&P

The secondary CLO markets: US vs. Europe The secondary market of US CLOs has grown dramatically in recent years. Total trading

volume of US CLOs in 2004 was estimated to be around $23-$25 billion, more than double

the amount in 2002.148 The growth was driven by several factors, including improving

transparency and liquidity, strong performance of most CLOs, and improving analytical

capabilities of investors. The bid from vehicles for BB and higher rated CLOs has been

very strong while B and lower rated paper, in addition to non-Moody’s rated tranches, is

largely trading to hedge funds and prop desks. We have even seen significant activity in

non-rated equity tranches.

Compared to US secondary market, European CLO secondary market is still in its infancy.

However, as the primary European CLO market develops further, the secondary market

should evolve similarly.

148 Based on estimates from CSFB trading desk.

31 March 2006


Summary After comparing the leveraged loan and CLO markets of US and Europe, we summarize

key observations and recommend the following:

1. Compared to the US leveraged loan market, the European leveraged loan market

is less efficient, as evidenced by its less responsive pricing system and lack of

rating coverage. However, this inefficiency actually makes European CLOs more

attractive, as higher liquidity premium could be passed along to CLO investors.

2. On the other hand, the inefficiency of the European loan market also makes it

more challenging to analyze European CLOs, which in turn becomes an

impediment for further improvement of liquidity and transparency.

3. We do not think there is substantial industry over-concentration in either US or

European CLO collateral pools.

4. We strongly believe that there are many opportunities in the European CLO

market. However, investors need to be mindful of some of the issues such as lack

of historical performance data and differences in the legal systems and

bankruptcy procedures of different jurisdictions. For investors unfamiliar with the

European market, we recommend investing in managed deals, leveraging a

manager’s expertise.

31 March 2006


Chapter 3. Trust Preferred CDOs

31 March 2006


Diversified Bank Trust Preferred CDOs -

Primer149

Executive Summary The diversified trust preferred CDO (DTP CDO), one of the newest CDO products, first

appeared in the CDO market in 2000. A trust preferred security qualifies as Tier 1 capital

for issuing banks, but unlike common equity, it does not dilute shareholders’ ownership

and also reduces tax costs for issuers. While trust preferreds have been a favored capital

option for larger financial institutions, they are usually too expensive for smaller banks to

issue on a stand-alone basis due to high transaction costs. By forming a consortium of

regional banks to issue DTP CDOs, underlying banks benefit from the economies of scale

and CDO investors enjoy regional diversification. We believe DTP CDOs offer small banks

a viable way to raise Tier 1 capital.

We believe that DTP CDOs will not only provide a cost-efficient capital solution for small

banks, but also are likely to deliver strong returns, should collateral banks continue to

sustain a low failure rate. Recent bank history demonstrates a low failure rate, sound bank

fundamentals, highly detailed disclosure, rigorous regulatory oversight, and better risk

management, all of which have resulted in increased public confidence in banks. Regional

banks, mostly small banks, not only share those positive attributes, but also focus mainly

on consumer finance and have far lower exposure to large corporations, which have been

the source of recent negative headlines. Regulators remain strong proponents of industry

consolidation, which reduces competition, improves efficiency and enhances profit

margins. We think small banks are well positioned in the event of consolidations and can

benefit from acquirers’ larger and higher credit quality franchises.

DTP CDOs present institutional investors an efficient way to gain exposure to diversified

pools of regional bank trust preferreds, a previously unavailable asset class with favorable

risk/reward characteristics. For eligible commercial banks, even including the 1980s’

banking crisis, historical bank failure statistics imply an average triple-B default rate, better

than most HY CBOs’ underlying credits, which are typically single-B rated. There is also

clear evidence that DTP CDOs offer regional diversification. We think well-capitalized

regional banks that have focused management teams, robust customer bases and strong

deposit franchises will perform as strong credits, enhancing DTP CDOs’ performance.

DTP CDOs offer long term investors seeking exposure to the banking sector attractive

relative value opportunities, as DTP CDO notes offer substantial spread pick-up over other

more established products such as HY CDOs, with greater credit enhancement and better

collateral credit quality.

In our view, primary risk factors associated with investing in DTP CDOs are uncertainties

related to long-term bank credit quality, adverse collateral selection issues, the lack of

diversification beyond the banking sector, longer average lives and the short history of

DTP CDO performance records.

149 This section was originally written by Neil McPherson, Helen Remeza, and David Kung, October 2003.

31 March 2006


ORIGIN, EVOLUTION AND FUTURE

Origin

Diversified trust preferred CDOs are backed by a pool of trust preferred securities. In

August 1996, the Federal Reserve Board (FRB) approved trust preferred securities as Tier

1 capital, which resulted in the trust preferred issuance boom.

Trust Preferred Security

A trust preferred security promises to make periodic coupon payments and has a stated

maturity (debt like), generally 30 years. Unlike debt, it is required to make the coupon

payment only when the issuer is financially able (equity like). Otherwise, interest may be

deferred for up to five-years, and the deferred interest is paid back on a cumulative basis.

A trust preferred security is a bullet bond (not an amortizer) with a 5 or 10-year non-call

period. After that, it is callable, usually at par (but not always). The “equity like” nature

enables a trust preferred security to be qualified as equity for regulatory capital purposes,

while its “debt like” nature enables the coupon payment on the security to be tax-

deductible for issuers, unlike other forms of equity.

Tier 1 Capital

Tier 1 capital, also known as core equity capital, is defined as the sum of common equity

and perpetual preferred stock, less any ineligible intangible assets. The FRB requires Tier

1 capital to constitute at least 50% of total capital, and trust preferred and perpetual

preferred stock to constitute at most 25% of Tier 1 capital.

From 1996 to 1998, a flood of $32 billion trust preferreds reached the market, as highly rated

financial institutions rushed to lock in the Tier 1 treated trust preferreds (Exhibit 162). The

trust preferred new issue market slowed down in 1999 and 2000, as the demand from

institutional investors slowed down. Most issuance since 2001 has been distributed through

retail investors. As of June 2002, public trust preferred issuance reached $55 billion in total.

Exhibit 162: Public Trust Preferred Issuance by Year and Issuer Asset Size*

0

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

1995 1996 1997 1998 1999 2000 2001 2002 (as of June

2002)

Iss

ua

nc

e (

bil

lio

n)

0

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

Iss

ua

nc

e (b

illion

)

<$200mm ($200mm, $1bn) ($1bn, $5bn) >$5bn other

Source: Credit Suisse, SNL

* This excludes the trust preferreds issued through DTP CDOs.

The FRB granted

trust preferreds as

Tier 1 capital in 1996

31 March 2006


To stay competitive amid the current pace of regulatory and market change, banks often

need to raise more capital. Using trust preferreds to raise capital is attractive for financial

institutions. Unlike common equity, though qualifying as Tier 1 capital, a trust preferred

security does not dilute common equity ownership; i.e., there is no dilution on voting rights,

earnings and return on equity. Also, the interest of trust preferreds is tax-deductible for the

issuers, making it a “cheaper” funding source.

The proceeds of trust preferred securities are typically used for a number of purposes by

the issuer, including:

1) Buying back common stock

Banks often use the trust preferred proceeds to retire the likely more “expensive”

common stock, while preserving regulatory capital requirements and enhancing

stock returns.

2) Funding product or service expansions

Bank customers increasingly demand better “one-stop shopping” capabilities

including access to multiple products, such as mutual funds and stocks. Better-

capitalized larger banks have the upper hand when expanding into new product

lines.

3) Financing a current or pending acquisition, or creating a cash and equity reserve

for future acquisitions

The repeal of Glass-Steagall prompted mega-mergers of commercial banks and

investment banks, allowing banks to cross-sell products, leverage economies of

scale and diversify product lines.

Evolution While larger banks have benefited from using trust preferred as a cheaper capital source,

regional banks had not prior to the advent of DTP CDOs. Regional banks typically are

smaller and specialize in consumer and small business lending across regional localities.

Since regional banks typically are small banks, we will use “regional bank” and “small

bank” interchangeably.

Historically the trust preferred utilization rate differed dramatically across banks (Exhibit

163). Excluding the trust preferreds issued through DTP CDOs and by bank count, we

estimate that only 2% of smaller banks issued trust preferreds vs. 32% for larger banks

(with more than $5bn in assets) vs. 60% for the largest banks (top 50 banks by asset size).

Exhibit 163: Public Bank Trust Preferred Issuance by Issuer Asset Size *

0

10

20

30

40

50

60

< $200 mm $200 mm and $1 bn $1 bn and $5 bn > $5 bn

To

tal Is

su

an

ce (

$ b

illi

on

s)

0

50

100

150

200

250

300

350

400

450

500

Issu

er C

ou

nt

Issuance total

Cumulative issuer count *

Source: Credit Suisse, SNL

* This excludes the trust preferreds issued through DTP CDOs.

Using trust

preferreds to raise

capital is attractive

for banks

Three common

reasons to issue

trust preferreds

Only 2% of small

banks issued trust

preferreds vs. 60%

for the top 50 largest

banks

31 March 2006


The issuance disparity between the market share of banks and trust preferreds is largely

attributable to high transaction costs to issuers and investor perceptions of greater event

risk and less liquidity associated with individual small banks. These factors impede small

banks’ ability to access the capital markets on a stand-alone basis.

DTP CDOs present a “win-win” solution for issuers and investors, because they enable the

economies of scale and regional diversification needed to make a deal viable. In a pooled

issue, the underwriting process is largely standardized and simplified across individual

banks. For example, a standard set of documents is used for all collateral banks, and no

independent road show or rating application is required for individual collateral banks. In

addition, DTP CDO investors are less sensitive to individual bank’s event risk as the pool

becomes more diversified. In Exhibit 164, we use two examples to illustrate how a DTP

CDO reduces transaction costs.

Examples

Suppose a bank with $300 mm in assets and $30 mm in capital desires to raise another

10% Tier 1 capital, or $3 mm, by issuing a trust preferred security. On a stand-alone basis,

the transaction costs (including underwriting, documentation, accounting, road show costs,

rating fee, legal fees) can be as high as $303,000, or 10.10%. In a pooled issuance, the

fee is around $128,100, or 4.27%. This amounts to a total saving of $174,900, or 5.83%, in

fees. In the second example, the issue size is $20 mm, larger than in the previous

example. The savings in this case is $302,400, or 1.51%. While in percentage terms the

savings is lower than the previous example, in dollar terms it is still quite a meaningful

saving to the trust preferred issuers.

Clearly, larger banks are more likely to execute stand-alone offerings, as their issue size

tends to be larger, where for them the savings over a pooled issuance is not as significant.

In addition, they may prefer the higher visibility from an independent offering, and, as such,

we believe the DTP CDO technology will mainly benefit smaller banks.

Cost and event risk

were the key

impediments

DTP CDOs present a

“win-win” solution

for banks and

investors

Reduction in

transaction costs is

substantial via DTP

CDOs

Exhibit 164: Examples of Trust Preferred Offerings: Savings in Transaction Costs

$3,000,000 $20,000,000

Offering Type

Pooled

Issuance

% of

Issuance

Pooled

Issuance

% of

Issuance

Pooled

Issuance

% of

Issuance

Pooled

Issuance

% of

Issuance

Placement/

Underwriters Fee 90,000 3.00% 112,500 3.75% 600,000 3.00% 750,000 3.75%

Legal Fees 30,000 1.00% 100,000 3.33% 30,000 0.15% 100,000 0.50%

Printing - - 20,000 0.67% - - 20,000 0.10%

Accounting - - 50,000 1.67% - - 50,000 0.25%

Trust Expense 5,100 0.17% 13,500 0.45% 5,100 0.03% 13,500 0.07%

Sub Total 125,100 4.17% 296,000 9.87% 635,100 3.18% 933,500 4.67%

Annual Trust Fees 3,000 0.10% 7,000 0.23% 3,000 0.02% 7,000 0.04%

Grand Total 128,100 4.27% 303,000 10.10% 638,100 3.20% 940,500 4.71%

First Year Savings $174,900 5.83% $302,400 1.51%


31 March 2006


Future

DTP CDOs have already changed the landscape of the trust preferred market. As of June

15th 2002, approximately 450 regional banks issued a total of $5.2 billion trust preferreds

through DTP CDOs.150 This almost doubled the trust preferred issuer base and increased

the size of the trust preferred market by about 10%.

We believe small banks will continue to use the DTP CDO platform to issue trust

preferreds, which are treated as the Tier 1 capital. The advent of DTP CDO marks a new

era where improving capital adequacy via trust preferred issuance is no longer only a

game that bigger banks can play. Small banks can play it well, too. This is a big step

towards establishing a level playing field for all banks. In fact, by deal count, year-over-

year by DTP CDO issuance increased by 100% and 50% for 2001 and 2002, respectively.

In all, DTP CDOs reached $12.9bn or 29 deals as of October 2003 (Exhibit 165).

Exhibit 165: DTP CDO Issuance (2000~October 2003)

0

0.5

1

1.5

2

2.5

3

3.5

4

4.5

5

2000 2001 2002 Oct-03

Issu

an

ce (

$ b

illi

on

)

0

5

10

15

20

25

30

35

Deal C

ou

nt

$ Annual Issuance

Cumulative DTP CDOs Issued (by Deal Count)


That said, we expect DTP CDO issuance to slow down slightly but nevertheless be steady.

The fast pace of DTP CDO issuance (since its inception in 2000), coupled with the current

stringent eligibility criteria (such as the 10% pro forma Tier 1 requirement) reduces the

availability of quality collateral banks.

We caution that should regulators challenge the tax-advantaged status of trust preferreds

as they did once before in 1997, there may be additional uncertainties surrounding trust

preferred issuance and, as such, DTP CDOs issuance may be affected. However, we

don’t foresee this in the near future, in light of the broader acceptance of trust preferreds

across US and Europe, and DTP CDOs’ increasing contribution to level the playing field

across the banking sector.

150

Among the 11 DTP CDOs priced as of June 15th 2002, there were a total of 350 collateral banks in

seven deals, while we estimated another 100 banks were represented in the other four deals in the pipeline at that time.

DTP CDOs almost

doubled the trust

preferred issuer

base

The advent of DTP

CDO levels the

playing field, in our

view

We expect the

issuance pace to

slow down

Caveat Emptor

31 March 2006


What’s Under the Hood: The Collateral As with other CDO products, it is important to understand the collateral - the trust

preferreds in the context of DTP CDOs. We focus on three aspects: collateral default rate,

bank selection criteria and surveillance. We will discuss key CDO structuring assumptions

such as regional diversification and structural enhancements in the next section.

Assessing the underlying collateral

Perhaps the greatest difficulty in analyzing DTP CDOs is in assessing collateral credit

worthiness. Collateral banks in DTP CDOs are often too small to be rated by the three

major rating agencies.

Rating agencies have approved two primary approaches to evaluate the collateral credit

quality for DTP CDOs. These include obtaining rating estimates for each issuer, or

adopting a “pooled approach”.

(1) Rating Estimate

For a fee, issuers can obtain estimated ratings for some or all of the individual trust

preferreds in DTP CDOs. Some issuers may apply a combination of the two approaches,

i.e., paying for the estimated ratings for selected banks but implementing the “pooled”

approach for the rest of the collateral banks.

(2) “Pooled” approach

The “pooled” approach assumes that collateral banks possess a similar credit quality as

the overall bank universe; i.e., collateral banks perform at the average of the overall bank

universe. This is a reasonable assumption if the sample size of the pool is relatively large.

DTP CDO portfolios often consist of a relatively large number of banks (i.e., ranging from

30 to 75 banks), and rating agencies have deemed the “pooled” approach applicable for

these portfolios.

We outline the pooled approach as follows:

1) Historic bank intervention statistics of the overall bank universe are used to infer

an average bank’s credit quality.

2) The credit quality of a trust preferred is assumed equivalent to that of the

issuing bank or bank holding company (BHC); i.e., a trust preferred will default

following the default of its issuing bank or BHC. The subordination nature of

trust preferreds is reflected in low recovery rates.

3) 3) Bank selection criteria are applied to eliminate “weaker” banks. By choosing

slightly larger banks with better capital adequacy and longer track records, we

believe a positive credit selection bias is established for DTP CDOs.

We based our study on FDIC’s Historical Statistics on Banking, which provides

comprehensive lists of individual banks that failed or received financial assistance from the

FDIC, collectively bank “interventions.”

We approximate the bank default rate by the intervention rate, and then estimate bank

credit quality by comparing the intervention rates to rating agencies’ benchmark corporate

default rates. For example, a one-year intervention rate of between 0.2% and 0.4%

indicates approximately a ‘Baa3’ rating. Appendix 1 lists Moody’s benchmark default rates.

Two ways to infer

collateral banks’

credit quality

Outline the “pooled”

approach

Using bank

“intervention” to

approximate bank

defaults

31 March 2006


The FDIC’s intervention rates are a conservative measure for bank failure rates or trust

preferred default rates, for at least two reasons:

4) All FDIC interventions were counted as defaults.151 An intervention does not

always necessarily imply a bank failure or the default of banks’ obligations

including trust preferreds. For example, an intervened bank might continue to

operate under some arrangements, enabling it to continue to meet partial or all

obligations.

5) The intervention rate is computed on an occurrence basis, and each bank was

counted individually; i.e., if one multi-bank holding company experienced five

defaulting subsidiaries instead of one, five defaults were counted.

To be concise, we will use “failure rate” consistently in the remaining text.

A Case of Bank Intervention

First City Bancorporation is an example of a failed bank that continued to partially meet its

obligations, thanks in part to effective regulatory oversight.

First City Bancorporation Inc., headquartered in Texas, was the fourth largest bank holding

company in Texas in 1988, with $11.2 billion in assets and 60 banking subsidiaries. First

City grew rapidly during the oil boom, but later suffered heavily due to the crisis in

agriculture, energy, and real estate markets. In 1987, First City approached the FDIC for

assistance. In 1988, the FDIC finalized the assistance plan that included injecting $500

million in new capital through a stock offering and transferring troubled assets to a

separate entity. Despite the FDIC resolution, First City’s asset quality continued to

deteriorate as losses mounted and the Texas economy sagged.

In 1992, concerned about the weakening First City rippling through its regional economy,

the FDIC stepped in once again. It created 20 bridge banks to assume deposits and some

assets and liabilities from the failed banks, and then proceeded to sell all of the bridge

banks in 1993. The acquiring institutions assumed all of the deposits and nearly all other

bridge bank liabilities. In 1994, the FDIC announced that all creditors with valid claims

were to be paid in full. In May 1995, more resolutions were announced including senior

preferred shareholders would be paid over two years; junior preferred shareholders would

receive between $100 million and $150 million and 35% of the new company’s common

stock; common shareholders would get 15% of the new company’s stock. Had First City

had trust preferreds outstanding, trust preferred holders would likely have been paid in full

or would have recovered considerably as they are typically pari passu to senior preferred

shareholders.

Historical bank failure rates The cumulative152 commercial bank failure rates across 31 years (1970~2002) suggest

that, even including the 1980s’ banking crisis, the commercial bank universe exhibited

lower failure rates than ‘Baa3’ corporates (Exhibit 166). For example, the inferred credit

quality of the overall commercial bank universe is better than most HY CBO collateral

quality, which is typically single-B rated.

151

Resolutions Handbook, FDIC. Types of FDIC intervention include assistant transactions, re-privatizations, re-openings, purchases or assumptions, insured deposit transfers, consignment program institution, and pay offs. Assistant transactions and pay offs are no longer used today. 152

This conclusion is based on cumulative failure rate (CFR), which is calculated by simply summing all annual failure rates. While we understand that ideally the CFR should be calculated based on a “cohort” study, the data required for the cohort is not available.

Intervention rate is a

conservative

measure for bank

failure rate

Historic bank failure

rates imply a triple-

B rating

31 March 2006


Exhibit 166: Cumulative Bank Failure Rate and Corporate Default Rate

0%

5%

10%

15%

20%

25%

1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002

Cu

mu

lati

ve

De

fau

lt R

ate

Commercial banks Baa3 rated corporate Baa2 rated corporate

Source: Credit Suisse, FDIC, Moody’s.

Though the “Baa3” implied commercial bank credit quality was drawn from the overall

universe, which includes larger banks, it should be applicable to small banks. This is

because the failure rates were weighted by the number of banks, and over 98% of banks

are small banks with asset sizes of less than $5 billion.

It is also worth noting that including the 1980s crisis in the rating estimation procedure

implies additional conservatism. Exhibit 167 contrasts the implied ratings from two different

timeframes: 1) the 1993~2002 period (which excludes the 1980s’ banking crisis), and 2)

the 1984~2002 period (which includes the 1980s’ banking crisis). The implied ratings are

derived from comparing the historical annual bank failure rate to Moody’s idealized annual

default rates.

Excluding the 1980s, the data suggest a lower failure rate and, therefore, higher credit quality.

As shown, the annual commercial bank failure rates imply an average credit quality anywhere

between single-A and triple-B, higher than the triple-B derived from the 1984~2002 data. This

trend of lower failure rates is also consistent across banks of different sizes. This suggests that

the banking system in general has been far healthier in the last decade than it was in the past

two decades.

The data excluding

the 1980s’ crisis

suggested a higher

bank credit quality

Exhibit 167: Annual Failure Rates for Commercial Banks*

1993-2002 1984-2002

Bank Asset Size Annual Failure Rate Implied**Credit Quality Annual Failure Rate Implied Credit Quality

Less than $200 mm 0.10% BBB 0.64% BBB

$200 mm to $500 mm 0.09% BBB 0.45% BBB

$500 mm to $1 bn 0.03% Higher than A 0.44% BBB

$1 bn to $3 bn 0.09% A 0.38% BBB

$3 bn or More 0.00% Higher than A 0.21% BBB

Weighted Average (WA) 0.10% A~BBB 0.61% Low BBB

WA excluding <$200 mm 0.07% High BBB 0.41% BBB


* Since the majority of collateral banks in most DTP CDOs are commercial banks and not S&Ls, we do not include S&Ls here.

** Derived from comparing to Moody’s idealized default rates

31 March 2006


Bank selection criteria

By choosing larger banks with better capital adequacy and more established track records,

a positive credit selection bias is established for DTP CDOs. Some typical eligibility

criteria for collateral bank inclusion are:

• Asset size greater than $200 mm

• Pro forma Tier 1 capital ratio greater than 10% 153

• Chartered five-years and longer

Let’s discuss each of these criteria in turn.

Small regional and community banks typically focus on consumer and small business and

stand to benefit from their in-depth local market knowledge and unequaled customer

relationships. Being close to their customers allows them to be more proactive in

managing problem credits, resulting in fewer loan losses. We believe that small bank

fundamentals remain sound, as they continue to maintain sufficient margins, manageable

asset quality and ample capital.

That said, for some of the smallest banks, the difficult banking environment of the 1980s

was challenging. These banks failed at a higher rate, largely as a result of concentrated

exposures to distressed commercial real estate loans, lower net interest margins from a

high interest rate environment and regional economy recessions. For example, in the

1980s, the cumulative failure rate of these smallest banks increased at a faster rate than

banks of other sizes (Exhibit 168).

Exhibit 168: Cumulative Failure Rates by Bank Size (1984~2002)

0%

2%

4%

6%

8%

10%

12%

1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002

Cu

mu

lative

Co

mm

eri

ca

l B

an

k F

ailu

re R

ate

Less than $200mm $200mm ~ $500mm $500mm ~ $1bn

$1 ~ $3bn $3bn or more

Source: Credit Suisse, FDIC

Trimming down exposure to these smallest banks reduces the historical bank failure rate.

For example, for banks with assets between $200 mm and $1 billion, the 17-year

(1984~2002) cumulative default rate drops from 12% to 8%, implying a pickup in credit

quality from ‘Baa3’ to ‘Baa2’, based on Moody’s idealized cumulative default rate

benchmarks.

153

The 10% is “pro forma”; i.e., the calculation includes the trust preferred issuance via DTP CDOs.

Size Matters

…but the smallest

banks suffered in

the 1980s

Reducing exposure

to the smallest

banks leads to a

pick-up in credit

quality

31 March 2006


Banks that failed had lower equity-to-asset ratios than surviving banks in the year before

recessions. These recessions include the 1982 agriculture recession in the Southwest, the

1990 real estate downturn in Northeast, and the 1991 California real estate depression.

By selecting banks with better capital adequacy, i.e., higher equity-to-asset capital ratios

or higher capital-to-asset ratios, we believe the collateral banks in DTP CDOs are less

likely to fail.154

Newly chartered banks failed with greater frequency than pre-existing banks in the 1980s.

For example, among all the institutions chartered in 1980-1990, 16.2% failed through 1994,

compared with a 7.6% failure rate for banks that were already in existence on Dec. 31,

1979. This was partly attributed to an influx of new banks and mutual conversions during

this period.155 Amidst the recessionary environment of the 1980s, newly chartered banks

began operating at a time when inexperience was a distinct liability. Also, the newly

converted mutuals reacted to the pressure of increasing earnings and meeting

shareholder expectations by aggressively expanding loan portfolios to leverage initial

capital positions, a strategy which eventually led to severe loan losses.

If history is a guide, by selecting larger, better capitalized and more established

banks, larger and better capitalized, we are likely to see fewer bank failures.

Surveillance and bank disclosure

Another important aspect of DTP CDOs is deal surveillance. Underpinning good

surveillance is deal/collateral transparency. We think that current bank disclosure is

excellent, and this allows DTP CDO investors to obtain timely performance information.

For example, banks regulators require all commercial banks to file a quarterly Call Report,

and it is mandatory for all S&L institutions to file a periodic Thrift Financial Report, both

containing a wealth of financial information.156 Regulators also conduct periodic on-site

examinations, including interviewing management, auditing financials, and revising credit

indicators such as the CAMEL ratings.157

Numerous financial ratios are available in the Call Reports. We think the following aspects

are important to monitor:

1) Credit performance: often measured by non-performing assets, loan loss reserves,

charge-offs, etc. 2) Capitalization: often measured by equity to assets, Tier 1 capital ratio, total risk

adjusted capital ratio, etc. 3) Profitability: often measured by return on assets, return on equity, profit margin, etc.

4) Funding mix: a strong retail/commercial deposit franchise is very valuable for regional banks.

154

Banks with capital-to-asset ratio greater than 10% are considered “well-capitalized” by BIS. 155

Mutual conversions refer to mutual savings banks that converted to the stock form of ownership. See “Understanding the Experience of Converted New England Savings Banks,” Jennifer Eccles and John Keefe, FDIC Banking Review 8, no. 1 (1995). 156

Please see http://www.ffiec.gov/reports.htm 157

This refers to capital, assets, management, earning and liquidity.

Better capitalized

banks failed less

often

Older is better

Quarterly Call

Reports

31 March 2006


Aside from the financials, there are also some helpful “composite” bank surveillance tools,

which predict or provide credit quality indications for individual banks. For example,

available resources158 might include 1) Thompson Financial’s bank ratings, and 2) the

“bank failure calculator” from the Office of the Comptroller of the Currency (OCC) or

FDIC’s CAMEL ratings, should they become publicly available.

We caution that the previously discussed three bank selection criteria provide a good

starting point for establishing a positive selection bias, but they are insufficient for a proper

and thorough evaluation of bank credit. We advocate a careful selection of collateral banks

to limit adverse selection issues.

Key ingredients for CDO structuring Having gained some insight of the collateral fundamentals, we now focus on portfolio risk

and CDO structural enhancements. We believe regional diversification dampens collateral

portfolio risk. We also rationalize structural assumptions such as recovery value and

prepayment, and round out with a discussion of some structural enhancements, cash flow

analyses and relative values.

East meets West and all places in between

We believe local economics and regional legislative differences are key drivers of regional

diversification. By selecting banks from various locations, regional diversity is introduced to

DTP CDOs. In DTP CDOs, geographical concentration guidelines limit single

bank/location/region exposure and diversify collateral pools, resulting in a better risk/return

profile.

Evidence of regional differences There is clear evidence of regional differences (with respect to failure timing and drivers)

within the banking sector. Exhibit 169 shows that regional bank failures did not peak at the

same time, and were more largely concentrated in the Southwest.

Exhibit 169: The Timing & Magnitude of Bank Failures Differed Across Regions *

0

10

20

30

40

50

60

70

80

1970

1972

1974

1976

1978

1980

1982

1984

1986

1988

1990

1992

1994

1996

1998

2000

2002

Nu

mb

er

of

Ban

ks

0

50

100

150

200

250

MidwestWestEastCentralSouthwest (right axis)


*Note Southwest is represented on the right scale

158 Off-site Surveillance Systems, FDIC.

Other “composite”

indicators

Caveat Emptor

Regional

diversification

The timing and

severity of bank

failures differed

across regions

31 March 2006


Clearly, factors affecting bank failures differed across localities, for example:159

• AL/LA/OK/TX/WY: severe economic downturns related to the collapse in energy

prices.

• CA/Northeast/Southwest: real estate related downturns.

• IO/KA/NE/OK/TX: the agricultural recession of the early 1980s.

• CA/TX: an influx of banks chartered in the 1980s and the parallel phenomenon of

mutual-to-stock conversions (MA).

• CO/IL/KS/TX/WY: regulation prohibited branching that limited banks’ ability to

diversify loan portfolios geographically and only allowed funding growth through

core deposits.

• NY/PA: the failure of a single large bank (Continental Illinois in May 1984) or a small

number of relatively large banks.

We can visualize the dispersion of bank failures by color-coding and shading the states

(Exhibit 170). For example, the West and Southwest region of the US experienced the

worst bank failure rates, while the Central and Southeast generally had less than 5% bank

failure from 1980~1994. Clearly, historical bank failures display regional patterns.

Exhibit 170: Historical Bank Failure Distribution Displayed Regional Patterns

(1980~1994)*

AK

LA

TX

OK

WY

AZ

CA

OR

HI

CT

UTCO

KS

NM

MA

NH

Cumulative Bank Failure Rate (1980-1994) State* Count Color Code

0% to 2% 12 3% to 5% 13

6% to 10% 9 11% to 15% 6

Over 15% 12


* Puerto Rico and US Virgin Island are included. The Appendix 1 provides FDIC’s six regional carve-outs.

159

“The Banking Crises of the 1980s and Early 1990s: Summary and Implications”, FDIC.

Regional economic

factors affecting

failure rates varied

Regional bank

failure patterns can

be depicted

31 March 2006


Quantify regional diversification To quantify collateral pool diversification, we implemented Moody’s alternative diversity

score framework to quantify regional diversity.160 While the agencies assume five regional

carve-outs, each of which is a separate industry category (just like in HY CBOs), the

beauty of the alternative approach is that one does not need to arbitrarily impose

“independent” regions. A key ingredient for the diversity score calculation is default

correlation for all the pair-wise collateral banks.

In analyzing bank failure history across the US, we observed that bank failure rates for

neighboring states tended to be more correlated than two distant states. We calculated

bank failure correlations between the states and show that, in fact, states within the same

region show higher failure correlation. Here, we applied the five-region carve-out used by

rating agencies (Appendix 1), and implemented Moody’s alternative diversity score

methodology, assuming all the trust preferreds in a DTP CDO share the same notional

and default rate.

Based on the FDIC’s bank failure data for the 50 states and the District of Columbia from

1966 to 2000, we construct a 51 by 51 correlation matrix using the method published in

the Journal of Fixed Income (see a summary of the approach in Appendix 2).161 The data

covers 13,529 banks in 1966 to 8297 banks in 2000, with 1,669 failures over 35 years

across the 51 locations.

For example, we estimated that two banks in the Southwest have a one-year intra-regional

correlation of 2.15%, while two banks in the Southwest and the East have a one-year

inter-regional correlation of 0.33%. To incorporate our view that default correlation may be

higher for longer terms, i.e., longer than a one-year horizon, we stress the default

correlations by five times to 10.73% and 1.63%, respectively. In our view, correlation is not

exactly “time” dependent, but it may vary across economic cycles; i.e., higher correlation in

market down cycles and lower correlation in bull markets. We view the five-times multiple

as reasonably conservative, but, of course, one may impose one’s own assumption.

Now, let’s look at a DTP CDO backed by 40 banks as an example. To obtain the diversity

score of the pool, we randomly pick 40 banks across 51 localities, which include the 50

states and the District of Columbia. The probability of the bank belonging to a specific

state is proportional to the number of institutions in that state. For example, since Texas

has a total of 710 (data as of the year 2000) commercial banks whereas Alaska only has

six, any randomly picked bank would be 771/6=118 times more likely to be in Texas than

Alaska. Across 100 randomly selected portfolios, the average diversity score is 20 and the

standard deviation is 2.0. We also re-ran the exercise using FDIC’s six-region carve-out

(Appendix 3), yielding a 22 diversity score and a 2.1 standard deviation.

Were the random bank selection rule to mimic reality (which we would not claim is always

the case) and if our default correlation assumptions are reasonable, we believe the typical

14 diversity score assignment to a well diversified 40-bank trust preferred CDO pool is

fairly conservative. All else equal, a conservative diversity score assumption results in a

more highly enhanced CDO structure.

160

“Moody’s Multi-sector CDO Rating Approach”, Moody’s, 2000. 161

“Default Correlation and Credit Analysis”, Douglas J. Lucas, Journal of Fixed Income, March 1995.

Apply Moody’s

alternative diversity

score

Quantify bank

failure correlation

Randomly sample

banks

A conservative

diversity score

results in a more

highly enhanced

CDO

31 March 2006


“Something about DTP CDO structuring” Aside from collateral default rates and regional diversification, other key structuring

assumptions include collateral recovery value, interest deferral frequency and prepayment.

We will round up this section by highlighting some structural enhancement features,

illustrating breakeven rates/multiples and comparing relative value.

Recovery value It is difficult to make any meaningful assessment of the recovery value of trust preferreds,

as very few troubled cases are known to us. This is partly due to the short history of the

bank trust preferred market and the healthy banking environment since the inception of the

trust preferred market in 1996. Nevertheless, after some “digging,” we identified Bay View

Capital Corp. (BVC) in California, as an example where the trust preferred has gone into a

“deferral mode.”

An example of trust peferred deferring interest

BVC is an example of trust preferred deferring interest. Barring any unforeseen

circumstances, we believe it is likely that BVC will make the trust preferred holders whole

eventually.

BVC is a bank holding company whose subsidiary is Bay View Bank, a retail and

commercial bank that operates 57 branches throughout the San Francisco bay area.

Originally “B1” rated by Moody’s, it issued $90 mm trust preferred in Feb. 1998. It acquired

FMAC, a franchise loan operation in 1999. Mainly due to the losses incurred by its

franchise loan lending, BVC had trouble maintaining healthy capital ratios, triggering the

FRB’s request of suspending its 9.76% dividend payment to its trust preferred holders

starting in Sept. 2000. Around that time, the bank’s rating was lowered to Caa1/D/CC by

Moody’s/S&P/Fitch. Subsequently, the bank completed its restructuring and raised over

$130 mm in new capital. In Oct. 2001, Moody’s upgraded BVC’s deposit and holding

company’s rating; in June 2002, Moody’s and S&P put all the ratings on positive watch,

citing that BVC has substantially improved its capital adequacy and is poised to return to

profitable operation.

Fitch assumes a 10% recovery rate on trust preferreds.162 While trust preferreds may

recover only a pittance in the event of failures due to their deeply subordinated nature, we

believe the 10% assumption is relatively conservative, in light of the following:

1) We use a broad bank failure or default definition; i.e., including all FDIC interventions

where in some instances, common shareholders were partially paid or eventually paid

in full. For example, in case study illustrated on Page9, had First City had trust

preferreds outstanding, trust preferred holders would likely have been paid in full or

would have recovered considerably as they are typically pari passu to senior preferred

shareholders and senior to common stock.

2) Preferred stock, which is typically junior to trust preferred, recovered 15% (median)

and 22% (mean) based on Moody’s 1970~2000 corporate data (including banks). We

also think with effective regulatory oversight (most other corporate sectors are not as

regulated), bank paper is likely to recover more than average corporates as bank

regulators are likely to proactively resolve bank failures and to restore public

confidence. This further supports the “conservatism” argument.163

162

“Bank trust preferred securities form new asset class for CDOs”, Fitch, February 16, 2001 163

“Default and Recovery Rates of Corporate Bond Issuers: 2000”, Moody’s, Feb. 2001.

Recovery value is a

conservative

“guess”

Two things point to

the “conservatism”

of a 10% recovery

rate used by Fitch

31 March 2006


Interest Deferral Trust preferred issuers have the ability to defer interest payment on trust preferred

securities for up to five years, though there is little evidence about how frequently and for

how long trust preferred security will defer payment. Rating agencies assume that deferral

will generally go with bank failures or regulator interventions, and thus will occur at the

same rate as default. Should deferred interest be paid back in full on a cumulative basis, it

is not an event of default. However, we think an event of interest deferral signals likely

credit troubles ahead.

Portfolio diversification can greatly reduce the impact of interest deferral. Simply put, for a

diversified portfolio, the likelihood of multiple banks missing interest payments at the same

time is not as likely. Also, compared to other CDOs, DTP CDOs have more excess spread,

which can be tapped to offset interest shortfalls.

Interest deferral does not usually affect PIKable164 mezzanine or junior CDO notes as

much as non-PIKable senior CDO notes, as timely payment of interest is not a must for

mezzanine or junior CDO notes. Interest deferral may affect senior noteholders’ ability to

receive timely cash flows. A typical mitigant is to establish a liquidity facility at the outset

and/or to fill it over time with available excess spread.

Prepayment After trust preferreds’ non-call period expires, collateral issuers may re-finance the paper

because: 1) they may lock in a low fixed rate in a lower interest rate environment; 2) as

issuers’ credit quality improves, they can borrow at lower rates, or 3) as the overall credit

spread of trust perferreds tightens due to the likely increase in demand for regional bank

paper, collateral banks can re-finance at lower rates.

Should collateral be called, static pool CDOs de-lever, likely resulting in higher

enhancement to CDO debt. This is almost surely a positive credit event for CDO debt,

conditional on no adverse selection in the collateral pool.

Consolidation has been prevalent in the financial sector in the 1990s. In fact, the drastic

40% decline of the total number of banks outstanding from over 14,000 in 1991 to fewer

than 10,000 in 2001 is a testimony of the rapid pace of M&A activity in the banking sector

(Exhibit 171). Interestingly, most of the decline is attributable to the buyout of banks with

less than $100 mm in assets, and more than 97% of banks being acquired since 1990 are

the ones with less than $5 billion in assets.

164

“PIKable” refers to the ability of the bond to “pay-in-kind” (PIK), which means deferred interest is paid back in accrued cash interest or more bonds.

Diversification and

excess spread

mitigate deferral risk

A liquidity facility is

often used too

Collateral being

called may result in

positive credit

events for CDO debt

31 March 2006


Exhibit 171: Number of Banks by Asset Size (1991~2001)

0

2,000

4,000

6,000

8,000

10,000

12,000

14,000

16,000

1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001

Nu

mb

er o

f B

an

ks

<$100m ($100m, $300m) ($300m, $1bn) ($1bn, $3bn) >$3bn


CREDIT SUISSE’s small-cap bank equity analyst, Lauren Lieberman, argues that in a

consolidating environment, small banks are in an enviable position.165 She also suggests a

couple of ways for small banks and investors to play the consolidation game. For example,

small banks are likely to be acquired by larger banks that seek to grow or to establish

platforms for new market expansion. Also, the banks may be consolidators themselves,

expanding product and distribution capabilities while eliminating back-office redundancies

and improving operational efficiency.

Despite a recent slowdown, we believe M&A activity will eventually resume as the pricing

power (i.e., equity valuation) of banks improves. Today, small banks are well positioned in

the event of consolidation and can benefit from larger and higher credit quality franchises

that can obtain more attractive financing, likely resulting in trust preferred prepayment.

On a cautious note, though, we think prepayment may lead to positive credit events for CDO notes, it also shortens the average life, likely resulting in lower total returns. To better understand the prepayment sensitivity, we examine how CDO cash flows vary across multiple prepayment assumptions in the next section.

Other elements

For DTP CDOs’ investors, it is also important to focus on 1) structural enhancement 2) cash flow analysis, and 3) relative value comparison. We will discuss these aspects in turn.

DTP CDOs have a long legal final (typically 30 years), and may have a longer average life

if collateral call rate is slow. Two common structural enhancement features seen in long

maturity CDOs are “debt turboing” and “an auction call,” both of which are intended to

reduce the average life of CDO debt.

Using excess spread to pay down the most expensive liability,166 usually the triple-B

rated tranche in a DTP CDO, can increase the amount of future excess cash flow and

shorten the triple-B average life. For tranches senior to the triple-B, credit enhancement is

not affected, as only excess interest (which would have otherwise been paid to equity) is

applied to pay down the triple-B, and this is conditional on the satisfaction of senior and

mezzanine coverage tests. In essence, subordination is “replaced” with OC (over-

collateralization).

165

“Small-Cap Banks – Three Ways to Play”, CSFB Regional Bank Research, June 6, 2002. 166

This is often conditional on the satisfaction of preset equity return targets.

Small-cap banks –

ways to play

Small banks poised

to benefit from M&A

Shortening the

average life of CDO

debt

Debt turboing

31 March 2006


Mandatory auction call redemption is another common feature with long maturity CDOs.

CDO trustees are required to conduct auction calls on a regular basis (they typically

coincide with payment dates, after the 10th anniversary of the transaction). To the extent

that the market value of CDO collateral pool is greater than the combined value of CDO

liabilities, the trustee should liquidate the collateral pool and use the proceeds to pay down

liabilities. Auction call redemptions are likely to enable an early return of principal, and, as

such, shorten the average life of CDO liabilities.

Barring any unexpected credit deterioration in the pool, it is likely that the auction call can

be exercised. At the auction call date, two occurrences are likely to have happened: 1) the

collateral would have seasoned and shortened its remaining average life, possibly being

sold at tighter spreads (or higher prices); 2) the triple-B would have been partly paid down

from debt turboing, reducing the amount of outstanding CDO liabilities. Both of these may

result in an in-the-money auction call; i.e., the value of the collateral pool being greater

than the value of the liabilities. Separately, after a CDO’s regular non-call period expires,

equity holders are increasingly likely to call the deal as the CDO may have de-levered

from triple-B turboing and collateral prepayment, which reduce the leverage and arbitrage.

We caution that should collateral credit deteriorate, both the auction call and the regular

call become less likely to be in-the-money. Central to CDO cash flow analyses are “breakeven” rates, which are the maximum collateral default rates before liability experiencing a first dollar loss (break in yield) or first principal loss (break in principal). Usually, the principal breakeven rate is higher than the yield breakeven rate, as the first loss of principal is a more severe scenario than the first dollar loss of cash flow.

As a generic example, we look at a DTP CDO with a $500 mm capitalization, with 54% in

‘AAA’, 21% in ‘AA’, 17% in ‘triple-B’ and 8% in equity. Exhibit 172 shows the breakeven

rates (see the footnotes of Exhibit 17 for modeling assumptions). For example, the triple-A

can sustain a 10.0% annual collateral default rate, or 64.2%, cumulative collateral default

rate before it begins to lose yield, and a15.1% annual, or 79.2% cumulative rate, before it

starts to lose a dollar in principal.

Another commonly used concept is “breakeven multiple,” derived from dividing breakeven

rates by a base case default rate. For DTP CDOs, we adopt Fitch’s 10% 30-year trust

preferred default assumption as the base case.168 Exhibit 173 shows the multiples for the

breakeven rates for the first dollar loss of principal, assuming that all collateral prepays at

year 10, 15 or 30. Clearly, early collateral prepayment results in a positive credit event for

CDO debt, enhancing the breakeven multiples. Interestingly, the equity multiples are quite

stable across different prepayment assumptions.

167

Cumulative default rate = 1 – (1- annual default rate/periodicity)^(periodicity*years of defaults). For example, for the ‘AAA’ tranche, 64.2%=1-(1-10.0%/2)^(2*10) for semi-annual pay trust preferred pools. 168

“Bank trust preferred securities form new asset class for CDOs”, Fitch, February 16, 2001.

Auction call

Breakeven analysis

Exhibit 172: Breakeven Rates for a Generic DTP CDO *

Based on a break in yield Based on a break in principal

Class

Annual

Default Rate

Cumulative

Default Rate 167

Annual

Default Rate

Cumulative

Default Rate

AAA 10.0% 64.2% 15.1% 79.2%

AA 7.0% 50.8% 8.9% 59.8%

BBB 3.2% 27.5% 5.0% 40.0%

Equity NA NA 2.3% 20.3%


* We assume 10-year bullet collateral, no deferral of interest payment on the collateral, constant default starting immediately, 10% recovery

with no lag, turbo ‘BBB’. Spread assumptions: collateral L+360bp, ‘AAA’ L+80bp, ‘AA’ L+110bp, ‘BBB’ L+375bp.

Sensitivity to

collateral

prepayment

31 March 2006


Exhibit 173: Principal Breakeven Multiples for a Generic DTP CDO

0

5

10

15

20

25

10 yr 15 yr 30 yr

Average Life of Bullet Collateral

Pri

ncip

al

Bre

akevn

Mu

ltip

le

AAA

AA

BBB

Equity


Conclusion The performance of DTP CDOs is directly tied to the health of the regional bank sector. In

our view, the overall banking sector possesses positive attributes such as low failure rates,

excellent disclosure, effective regulatory oversight and strong public confidence. Being

close to their customers allows small banks to be more proactive in managing problem

credits. We believe that regional banks’ fundamentals remain sound, as they continue to

maintain sufficient margin, manageable asset quality and ample capital, all of which

contribute to a low failure rate. Furthermore, regional banks are well poised in an

environment of consolidation and pools of regional bank trust preferreds offer geographical

diversification; both can be beneficial to DTP CDOs. We believe well-capitalized regional

banks that have focused management teams, robust customer bases and strong deposit

franchises will perform as strong credits, enhancing DTP CDOs’ performance.

Nevertheless, in our view, primary risk factors associated with investing in DTP CDOs are uncertainties related to long term bank credit quality, adverse collateral selection issues, the lack of diversification beyond the banking sector, longer average lives and the short history of DTP CDO performance records.

Should regional banks continue to sustain a low failure rate, we believe DTP CDOs not

only provide a cost-efficient capital solution for small banks, but also are likely to deliver

strong returns for CDO investors, a “win-win” for market participants. DTP CDOs offer

long term investors seeking exposure to the banking sector attractive relative value

opportunities, as DTP CDO notes offer substantial spread pick-up over other more

established products such as HY CDOs, with greater credit enhancement and better

collateral credit quality.

31 March 2006


Appendix 1. Rating Agency’s Five-region Carve-Out

West Midwest Central Southwest East

WA ID WI NM ME

OR MT MI TX NH

CA ND IL OK VT

HI MN IN AR MA

AK WY OH LA RI

SD KY CT

IA TN NY

NV MS NJ

UT AL PA

CO WV

NE MD

KS DE

MO DC

AZ VA

NC

SC

GA

FL

* This map displays Fitch’s five regional carve-outs. Each of the regions is colored and shaded, and the states included in the regions are listed

in the table.

Source: Fitch, Credit Suisse

31 March 2006


Appendix 2. FDIC’s Six-Region Carve-Out

NortheastRegion

Southeast Region

Central Region

Midwest Region

Southwest Region

West Region

Connecticut Alabama Illinois Iowa Arkansas Alaska

Maine Florida Kentucky Missouri Oklahoma Arizona

New Hampshire Georgia Ohio North Dakota Louisiana Montana

Pennsylvania Mississippi Indiana Kansas Texas California

Delaware North Carolina Michigan Nebraska New Mexico Colorado

Maryland South Carolina Wisconsin South Dakota Hawaii

New Jersey Tennessee Minnesota Utah

Rhode Island Virginia Idaho

Massachusetts West Virginia Nevada

New York Washington

Vermont Oregon

Wyoming

* This map displays FDIC’s regional carve-outs. Each of the regions is colored and shaded, and the states included in the regions are listed

in the table.


31 March 2006


An Introduction to Insurance Trust Preferred

CDOs169

Since the first insurance trust preferred (ITP) CDO was brought to the market in November

2002, the number of deals backed by ITPs and surplus notes has been growing gradually,

with five deals priced to date, bringing total outstanding ITP CDOs to $1.6bn (Appendix 1).

Proceeds from insurance trust preferreds and surplus notes issuance are often used for

financing acquisitions, funding company growth or demutualization, and for replacing

capital reduction from investment losses, etc.170 Just like pooled bank deals, we believe

ITP CDOs offer a more level playing field for smaller insurers (vs. larger ones), as the ITP

CDO platform allows smaller insurers to achieve lower financing costs. For example, trust

preferreds and surplus notes remain qualified for partial equity credit for rating agency

treatment, while other forms of debt such as bonds and loans do not. Further, the

application of CDO technology creates a “win-win” in that it also provides an opportunity

for mainstream fixed income investors to buy pooled insurance trust preferred risk with

product line and geographic diversity at an attractive spread.

To illustrate how an ITP CDO works, we’ll focus on:

• Collateral selection;

• Portfolio diversification;

• Collateral credit performance;

• Regulation and disclosure;

• The insurance sector outlook; and,

• Some unique structural enhancement features for ITP CDOs.

Collateral Collateral type

On average, outstanding ITP CDOs usually have about 35 insurers participating in the

program, with 70% in the property & casualty (P&C) sector and the balance in life & health

(L&H) companies (Exhibit 174) and others, roughly proportional to the market share (by

company count) breakdown between P&C and L&H companies.

ITP CDOs typically contain two types of collateral: trust preferreds and surplus notes.

Trust preferreds are issued by the holding company of a publicly owned company (“stock

company”). For an overview of the trust preferred issuance structure, please see

Appendix 2. Surplus notes are typically issued by a mutual company’s operating entity. A

mutual company is owned by policyholders rather than by shareholders. The largest US

P&C insurer, State Farm, is a mutual company. In 2001, mutual P&C insurers

represented 33% of industry surplus171, 27% of assets, and 32% of underwritings. Since

1978, the number of mutuals has dropped below that of stock companies and remained

fewer, partly because of the difficulty in raising capital and the absence of stock-related

incentives for management.172

169

This report was originally written by Neil McPherson, Helen Remeza, David Kung and Eric Zhai, December 3, 2003. 170

Demutualization refers to the event in which a mutual insurer is restructured as a stock company, which is then owned by shareholders rather than policyholders. 171

Surplus, or policyholders’ surplus, refers to the sum remaining after all liabilities are deducted from all assets. Essentially, this is an insurer’s statutory net worth. Surplus, in addition to loss reserves, provides financial protection to policyholders in the event that a company suffers an unexpected or catastrophic loss. 172

2003 Property-casualty insurance primer – 18th edition, February 2003, CSFB P&C insurance equity

research.

Five ITP CDOs

priced so far

ITP CDO platform

offers a “win-win”

About a 7:3 split

between P&C and

L&H insurers

Two main types of

collateral: trust

preferreds and

surplus notes

31 March 2006


So far, about 62% of the assets in ITP CDOs have been invested in trust preferred

securities issued by insurance holding companies, while 33% are invested in surplus notes

issued by mutual insurance operating companies (Exhibit 174).

Exhibit 174: Select collateral statistics for outstanding ITP CDOs

Average Range

# Financial Institutions 35 (31, 40)

Property & Casualty 72% (67%, 76%)

Other (Incl. Life & Health) 27% (19%, 33%)

Trust Preferred 62% (51%, 70%)

Surplus Note 33% (30%, 37%)

Source: Credit Suisse, Fitch.

A comparison of the two main types of securities found in ITP CDOs is illustrated below

(Exhibit 175).

Exhibit 175: A comparison between trust preferred securities and surplus notes

Trust Preferred Surplus Note

Key Characteristics * 30-yr maturity* Subordinated to other debt but senior to common and

preferred equity

* Regulators have oversight of the overall health of an insurer, for

which they may exercise their discretion to suspend dividends to the

holding company, which can result in deferring dividend payments to

trust preferred holders

* Unpaid dividends accrue until paid

* No voting or equity conversion rights

* 30-yr maturity, but possibly shorter

* Subordinated to secured debt and policy holders

* Interest and principal payments subject to prior state regulatory

approval (in the issuer’s domicile)

* No voting or equity conversion rights

Key attractions for insurers * Dividends paid are tax deductible to issuer

* Rating agencies give some equity credit for trust preferreds, resulting

in a lower leverage ratio for the insurance company

* Interest is tax deductible to issuer

* Mutuals can issue surplus notes, being treated as equity capital by

regulators, without having to demutualize

* Surplus notes generally go to increase issuer surplus, resulting in a

lower leverage ratio for the insurance company

* Rating agencies give some equity credit for surplus notes, resulting

in a lower leverage ratio

Market size As of 2003Q2, SNL data source indicates there are about $15bn of

trust preferred securities across 228 insurers outstanding in US

Since 1993 and as of Dec. 2002, $19.7bn surplus notes were issued

across 337 insurance companies, including $6.1bn in the P&C sector

and $13.6bn in the L&H sector


Collateral Selection

Similar to the banking sector, in which over 95% of banks have assets less than $10bn,

most companies in the insurance industry are relatively small. In 2003, there were 2,671

P&C and 1,164 L&H companies in the U.S. Most are very small, i.e., 88% (by company

count) with assets less than $1bn and 98% with assets less than $10bn. These smaller

insurers are taking advantage of the CDO platform, i.e., gaining more efficient funding.

This greater funding efficiency (and better access to the capital market) results from a

reduced issuance cost due to ITP CDO diversification, documentation standardization and

reduced road show requirements.

To provide some color on the underlying insurance companies in ITP CDOs, we list some

key insurer eligibility criteria as described by Fitch (Exhibit 176), i.e., insurers satisfying the

criteria shown below can participate in the pooled (CDO) program. Typically an eligible

pool can achieve an average Single-B rating equivalent by the rating agencies. While

having a snapshot of these indicators is helpful, we think investors should also keep a

close watch on evolving trends, which may shed additional light on creditworthiness.

Small insurers, a

large part of the

insurance universe,

benefit most from

ITP CDOs

31 March 2006


Exhibit 176: Fitch’s key eligibility criteria for insurers participating ITP CDO programs

P&C Companies Life Companies

1. Minimum five-year operating history Same

2. No significant businesses other than the core P&C insurance operation No significant businesses other than the core life insurance operation

3. Minimum of $30 million of statutory capital Same

4. Minimum NAIC risk-based capital (RBC) ratio of 150% Minimum NAIC risk-based capital ratio of 250% (under the new risk based capital ratio formula)

5. Maximum NPW/PHS of 2.5X NA

6. Maximum net leverage of 5.0X NA

7. Reinsurance Recoverables /PHS < 100% NA

8. Minimum 5-year return on surplus of 5% Same

9. Maximum risky assets to adjusted surplus ratio of 100% Same

10. Trust preferred capital represents less than 25% of total GAAP capital

structure

Same

11. Maximum adjusted leverage ratio of 45% on a GAAP basis Same

Source: Fitch, Credit Suisse

It is also common that the rating agencies may choose to obtain a "credit estimate" for

each underlying insurer in ITP CDOs. This score is typically assigned by the agencies’

insurance group and/or is derived from a credit-scoring model. Fitch suggests that it will

evaluate insurers using a model that considers factors such as capital/reserve adequacy,

profitability, investment allocation and risks, operating/financial leverage, and credit

exposures to re-insurers.

Portfolio Diversification As in all CDOs, collateral risk diversification results in more stable portfolio defaults/losses,

benefiting ITP CDO debt performance. Several aspects of ITP CDO collateral

diversification include diversification across sector, product line and geographic region.

Insurance sector classification

Based on the type of risk insured, broadly speaking, there are two types of insurance

companies:

• Property/Casualty (P&C); and,

• Life/Health (L&H).

Aside from the above, reinsurers and alternative insurers173 also offer risk coverage.

Product line classification

Insurance companies typically offer multiple product line coverage. For example, P&C

companies typically offer:

• Personal lines: auto, homeowners’ multiple peril174; and,

• Commercial lines: workers’ compensation, commercial multi peril, property

reinsurance, directors’ & officers’ liability, etc.

In 2001, total net premiums written for the P&C insurance industry were $329bn, 49.6%

and 46.1% for personal and commercial lines, respectively, according to A. M. Best.

173

Alternative insurance is often offered by industry or labor groups rather than an insurance company. In the alternative risk transfer market, the insured typically assumes a substantial amount of its own loss exposure, primarily the predictable, frequent losses, and transfers the less predictable, excess risks to insurers and reinsurers. In most cases, the client purchases unbundled services that include risk management, loss and claims control, and investment management. The benefits of these alternatives include lower and more stable insurance costs, greater control over the client’s risk management, and an increased emphasis within the client's organization on loss prevention and control. 174

Commercial Multi-Peril provides a wide range of coverage for commercial establishments, including property coverage.

P&C insurers

31 March 2006


Life companies typically offer the following type of policies:

• Whole life: universal, variable and universal variable life;

• Term insurance;

• Group life, accident & health insurance;

• Annuity;

• Other policies, such as credit insurance175 and industrial life insurance.176

Health insurers cover disability and supplemental health, etc.

It is important to examine ITP CDO pools by product lines, as companies within the same

broad sector (i.e., P&C, L&H) may have very different products. To illustrate this, we use

an example in a CREDIT SUISSE insurance credit research publication entitled,

“Hurricane Isabel: Could Create Some Buying Opportunities“.177 Exhibit 177 provides an

example of some P&C companies and their product line distribution by premium collection.

For instance, while the homeowners’ insurance exposure across 30 selected companies in

this example ranges from 0% to 24%, auto exposure extends from 9% to 90%. This

suggests that insurance companies operating in broad industry categories may have a

substantially difference product mix (or portfolio risk). Thus, examining an ITP CDO

portfolio diversification only by broad insurance sector classification (such as P&C or L&H)

may not offer enough detail. Aggregating premium collection across underlying issuers by

product line can shed additional insight.

Typically, sector and product line concentration are monitored and controlled, as rating

agencies often impose issuer, sector and product line concentration limits.

175

Term life insurance designed to cover the repayment of a loan, installment purchase, or other financial obligation. 176

Also know as home service life insurance, the premium is collected by the salesperson at the home of the insured on a weekly or monthly basis. 177

“Hurricane Isabel: Could Create Some Buying Opportunities”, CSFB, September 15, 2003.

L&H insurers

Product line

diversification

sheds additional

insight

31 March 2006


Exhibit 177: Insurer’s product focus differs P&C Companies Focused in States in Path of Hurricane Isabel, with At Least 20% of Premiums in Auto, Homeowners’ Commercial Multi-Peril and Property

Reinsurance Lines. Sorted by percentage of total net 2002 premiums written in New York, New Jersey, North Carolina, Virginia, Pennsylvania and Connecticut.

Figures in $millions unless indicated otherwise

Comm'l Total NPW

12 mo. 2002 Home- Multi- Property Focus in

P&C Group Total NPW Auto owners Peril [2] Reins Total Isabel States

1 Selective Insurance Group Inc 1,075.1 21% 3% 4% 0% 27% 73.5%

2 PMA Capital Insurance Group 1,054.2 9% 2% 5% 10% 27% 69.3%

3 Erie Insurance Group 3,330.0 52% 15% 11% 4% 83% 66.1%

4 Harleysville Insurance 1,126.3 22% 9% 28% 0% 59% 61.7%

5 White Mountains Insurance Grp 3,112.4 36% 9% 11% 1% 57% 36.5%

6 Amica Mutual Group 1,156.2 73% 22% 0% 0% 95% 34.4%

7 Travelers PC Pool 11,882.3 27% 12% 17% 0% 57% 33.3%

8 Ohio Casualty Group 1,448.6 26% 10% 20% 0% 56% 32.5%

9 Chubb Group of Insurance Cos 7,811.3 8% 14% 17% 1% 40% 31.9%

10 Berkshire Hathaway Ins Group [1] 15,203.8 46% 0% 3% 11% 60% 30.5%

11 Nationwide Group 11,740.5 56% 15% 9% 0% 81% 30.2%

12 MetLife Auto & Home Group [1] 2,876.5 74% 22% 0% 0% 96% 29.0%

13 Kemper Insurance Companies 2,089.6 14% 5% 10% 0% 28% 28.9%

14 Liberty Mutual Insurance Cos 10,573.6 34% 8% 8% 1% 51% 27.3%

15 Allstate Insurance Group [1] 23,342.1 71% 23% 2% 0% 96% 25.6%

16 Hartford Insurance Group [1] 8,394.7 29% 8% 18% 3% 57% 25.5%

17 Royal & SunAlliance USA 3,123.1 21% 4% 11% 0% 37% 25.1%

18 GMAC Insurance Group 2,634.5 56% 0% 0% 1% 58% 24.9%

19 Fairfax Financial (US) Group 3,003.0 9% 2% 7% 6% 25% 24.3%

20 Allmerica Prop & Casualty Cos [1] 2,269.2 51% 16% 15% 0% 81% 21.5%

21 Great American P&C Ins Group 2,386.8 36% 0% 6% 0% 42% 20.5%

22 American International Grp Inc [1] 21,045.8 19% 1% 2% 1% 22% 20.1%

23 USAA Group 6,967.0 75% 21% 0% 0% 96% 20.0%

24 Progressive Insurance Group 9,455.6 90% 0% 0% 0% 91% 19.8%

25 Allianz of America, Inc 2,631.9 18% 17% 23% 0% 58% 19.8%

26 Horace Mann Insurance Group [1] 523.1 72% 24% 0% 0% 96% 16.2%

27 State Farm Group [1] 42,747.4 67% 22% 2% 0% 92% 14.6%

28 Cincinnati Insurance Cos [1] 2,612.7 21% 9% 26% 0% 55% 14.0%

29 Westfield Group 1,252.0 31% 12% 19% 2% 64% 13.2%

30 Sentry Insurance Group 1,552.9 46% 2% 1% 0% 49% 12.0%

1) These companies all have sizable life insurance operations, premiums for which are not included in the above table. Life premiums as a percent of consolidated company premium are the

following – Berkshire Hathaway Inc. (3%), MetLife (91%), Allstate Corp. (9%), Allmerica Financial Corp. (15%), American International Group (52%), Horace Mann Group (15%), Cincinnati

Insurance Group (3%), State Farm Group (7%).

2) Commercial Multi-Peril provides a wide range of coverage for commercial establishments, including property coverage.

Source: Company Reports, AM Best, Credit Suisse

Smaller companies such as Midland Company (with market capitalization of $395mm as of

Sept. 2003) have a very different product profile than larger insurers (Exhibit 178). For

example, Midland specializes in writing physical damage insurance and related coverage

on manufactured housing, homeowners, lower valued homes, dwelling fire, mortgage fire,

collateral protection, watercraft and related insurance, segments on which larger insurers

may not focus.178

For pooled deals, small insurers’ niche product focus offers diversification. Of course,

smaller companies should be aware of excessive expansion in a competitive market, i.e.,

they should not write too much business that appears profitable in the short term when the

long-term prospect is questionable.

178

“Blame it on Isabel… and motor sports,” CSFB, P&C insurance equity research, September 2003.

Smaller insurers are

often niche

players…

31 March 2006


Exhibit 178. Small insurers such as Midland have a different product profile

than larger ones

Smaller players such as Midland Company (with market capitalization of $395mm as of Sept. 2003)

have a very different product profile than larger insurers such as Allstate. The gross premiums written

in 2002 (totaled $588 mm) can be broken down below respectively. While Midland (top chart) has

larger exposure in manufactured homes and motor sport but very little in auto, larger P&C insurers

provide more risk coverage to auto (about 38% as shown in the bottom chart).179

Midland Comapny

Watercraft

3%Mortgage Fire

3%

Recreational

Vehicle

3%

Commercial

Property

3%

Long Haul Truck

2%Other

3%

Collateral

Protection

4%

Credit Life &

Related

8%

Motor Sport

11%

Site Built Dwelling

13%

Manufactured

Homes

47%

Larger P&C

Other

15%Reinsurance

4%

Accident & health

5%

Commercial auto

5%

Liability other than

auto

7%

Commercial

multiple peril

7%

Workers'

compensation

8%

Homeowners

11%

Private passenger

auto physical

damage

16%

Private passenger

auto liability

22%

Source: Company Reports, Credit Suisse, S&P.

179

This is based on “Industry surveys – P&C Insurance,” S&P, July 17, 2003. The market share by product line is based on net premiums written and averaging over the A. M. Best P&C coverage universe. We think this is a reasonable proxy for larger insurers’ product profile, as the larger companies dominate the overall market.

31 March 2006


Geographical Diversification

There is some evidence that average insurer failure rates differ across the state of

domicile.180 While Wyoming, Louisiana, Montana, Puerto Rico and Florida experienced the

highest insolvency rate, Connecticut, Idaho, Kansas, Mississippi, New Hampshire and

North Dakota and the District of Columbia did not have any insolvency over the period

studied by A. M. Best (1969-1990).

To some extent, the difference in insolvency rates is probably related to the effectiveness

of state regulation, including licensing, regulatory and capital requirements, and the rate-

setting mechanism, which vary among the states. Different state requirements make it

easier to obtain licenses in some states than others.

Further, premium rates are regulated by individual states. There are two basic types of

rate-setting regulation: use and file (competitive rate), 181 and prior approval (by state

regulator). In jurisdictions where use and file is the law, rates are set on a competitive

basis and are subject to later audit. In the US, states where prior approval is the law

include California and the most heavily populated states in the Northeast, totaling 63.2%

by population today (Exhibit 179).182

Because insurer performance can differ across states, combining insurance risk

exposures in various states likely enhances CDO diversification. In addition, smaller

insurers tend to be more localized in their risk coverage, partly due to their regional

focus/knowledge and their ability to fill the gaps for larger insurers. For example, the

largest P&C company in the US, State Farm, recently reduced voluntarily its insurance

sales in certain states, offering opportunities for regional insurers (Exhibit 180).183

180

A. M. Best’s P&C Insolvency Study 1969~1990. 181

Essentially, an insurer establishes and uses its rate prior to filing them with a state regulator who may then challenge and ultimately change them. 182

“Allstate corporate,” April 14 2003, P&C Insurance, Equity Research, CSFB. 183

2003 Property-casualty insurance primer – 18th edition, Page 123, CSFB, P&C insurance equity research, February 2003.

Insurer failure rates

differ across

states…

…Partly driven by

the differences in

state regulation

There are two basic

types of rate-setting

regulation

Exhibit 179: Rate regulation by state

Arizona 1.8% Michigan 3.6% Vermont 0.2% Alabama 1.6% Maine 0.5% Oklahoma 1.2%

Arkansas 0.9% Minnesota 1.8% Virginia 2.5% Alaska 0.2% Massachusetts 2.3% Pennsylvania 4.4%

Colorado 1.5% Missouri 2.0% Wisconsin 1.9% California 12.2% Mississippi 1.0% South Carolina 1.4%

Connecticut 1.2% Montana 0.3% Wyoming 0.2% Delaware 0.3% Nebraska 0.6% Tennessee 2.0%

Idaho 0.5% Nevada 0.7% Dist. of Columbia 0.2% New Hampshire 0.4% Texas 7.4%

Illinois 4.4% Ohio 4.1% Florida 5.5% New Jersey 3.0% Washington 2.1%

Indiana 2.2% Oregon 1.2% Georgia 2.9% New Mexico 0.6% West Virginia 0.7%

Iowa 1.1% Rhode Island 0.4% Hawaii 0.4% New York 6.7%

Kentucky 1.5% South Dakota 0.3% Kansas 1.0% North Carolina 2.8%

Maryland 1.9% Utah 0.8% Louisiana 1.6% North Dakota 0.2%

% of Total US Population 36.8% 63.2%

Competitive Rate Prior Approval

The largest states in the United States from a population standpoint require prior approval.

Source: Insurance Service Office (ISO)

Combining small

insurers domiciled

in various states

enhances CDO

diversification

31 March 2006


Exhibit 180. Voluntary reduction in sales from larger insurers offers opportunities

for smaller regional companies

After a tough 2001, when State Farm (the largest US insurer) lost $75 on every car insured after tax and

investment income, it took some corrective actions. These include higher rates and a moratorium

(suspension of new sales) or limit on new homeowners’ sales in 26 states and on personal auto sales in

3 states (see the chart below). The largest of these states include Texas, California, and Florida, which

collectively represent an estimated 28% of the total U.S. homeowners’ insurance market and 25% of the

personal auto insurance market. In addition, the company announced a restructuring effort in August

2001 designed to consolidate its operations into 13 regions from 25.

While these measures are designed to improve State Farm's profitability, these limitations on insurance

sales provide opportunities for other players including smaller regional insurers.

ALASKA

CALIFORNIA

IDAHO

OREGON

MONTANA

WYOMING

UTAH COLORADO

ARIZONA

NEW MEXICO

TEXAS

OKLAHOMA

KANSAS

NEBRASKA

SOUTH DAKOTA

NORTH DAKOTAMINNESOTA

WISCONSIN

IOWA

ILLINOIS OHIOIN

KENTUCKY

WV

VIRGINIA

NO.

CAROLINA

GEORGIA

FL

ALABAMA

MS

MISSOURI

ARKANSAS

LA

NEVADA

HAWAII

MICHIGAN

PENNSYLVANIA

NJ

NEW YORK

CT

MA

VT

NH

MAINE

TENNESSEE

CAROLINA

SO.

DE

RI

DC

Moratorium - Homeowners

Limits on Sales - Homeowners Limits on Sales – Auto & Homeowners

Moratorium – Auto & Homeowners

COLORADO

NEW MEXICO

ILLINOIS

IDAHO

OREGON

WASHINGTON

MONTANA

WYOMING

UTAH

ARIZONA

KANSAS

NEBRASKA

SOUTH DAKOTA

NORTH DAKOTA MINNESOTA

WISCONSIN

IOWA

OHIOIN

MISSOURI

ARKANSAS

NEVADA

TENNESSEE

MD

PA

MI

WV

Does Not Operate State Farm insures roughly one over every

five cars and homes in the United States.

Source: Credit Suisse,

31 March 2006


Insurance Company Default Studies

Data universe

Data presented in this paper are mainly based on A.M. Best’s data. A.M. Best

(www.ambest.com), located in New Jersey, has been the leading tracker of the insurance

industry for over 92 years. It covers over 3,835 P&C and L&H companies (including 2,671

P&C companies), which represent 99% of the domestic industry’s assets and premium

volume.

There are some discrepancies between industry convention and the rating agencies’

definition of default. For example, insurance industry convention for P&C failures is

insolvency and for L&H the standard is financial impairment, while rating agencies’

definition of debt default is often payment impairment.

We think defaults as defined by the insurance industry (used in A.M. Best’s insolvency and

impairment study) can lead to a conservative measure for trust preferred and surplus note

defaults, resulting in a higher number of defaults than rating agency definitions may have

indicated. This likely over-counting is due to the fact while insolvencies or financial failures

are recorded for individual companies and/or multiple times (if a company failed, re-

established itself and failed again), trust preferreds are issued at the holding company

level. For example, Central National Insurance Co. of Omaha and of Puerto Rico failed in

1989 and 1991 (but the holding company survived), respectively, and were counted as two

events by A.M. Best, while Commercial Standard Insurance Co. failed twice, in 1981 and

1985, which were recorded as two insolvencies by A.M. Best. 184 Multiple company

insolvencies or financial failures under the same holding company thus could overstate the

total number of trust preferred defaults.185

Historical failure statistics

Having noted the differences between rating agency defaults and industry conventions, for

simplicity, we address insolvency and financial impairment generically as "default". Exhibit

181 summarizes insurers’ historical default experiences.

Exhibit 181: Summary of historical insurer default experiences

Annual Average Default Rate Standard Deviation Time frame

P&C 0.94% across 22 years, indicating a Ba1 credit* 0.55% 1981~2002

L&H 0.48% across 26 years, indicating a low Baa3 credit* 0.45% 1976~2001

* Indicated ratings are based on Moody’s idealized annual default rates

Source: A.M. Best, S&P and Credit Suisse

We offer the caveat that ideally these statistics should be re-evaluated, tailored to smaller

insurers, the main participants of ITP CDO programs, and adjusted to the portfolio makeup

of stock companies vs. mutuals.186 However, this task is formidable at this time due to

limited data availability.

184

“Excess and surplus 2003,” special report, September 2003, A.M. Best. 185

In addition, theoretically, insolvency is a much broader concept than payment impairment. For the purpose of A. M. Best’s study, the insolvency count includes any U.S. domiciled insurance company against which action has been taken by the insurance department in its state of domicile for reasons of financial impairment. State actions include administrative orders, supervision, receivership, conservatorship, liquidation or another form of action, which restricts or limits an insurance company’s freedom to conduct business. This is likely to be more extensive than payment impairment, which may result in the over-counting. 186

A. M. Best noted a significant greater insolvency rate for stock companies than mutuals in its study covering the 1969~1990 period.

A likely source of

over-counting

31 March 2006


P&C companies

P&C insolvency experiences can be summarized as follows:

• The annual P&C company insolvency rate averaged 0.94% (with a standard deviation of

0.55%) across the 22 years. Roughly, this implies a 9.4% (=0.94%*10) 10-year default

rate, assuming a constant annual default rate of 0.94%, which implies a high Double-B

rating based on Moody’s corporate default rate.

• Companies rated in the C and C- categories by A. M. Best (Appendix 2 illustrates A. M.

Best's rating scale) experienced the highest insolvency rate three years later (Exhibit

182). Eliminating companies with low ratings, i.e., based on the eligibility criteria

discussed before, from an ITP CDO pool should lead to more favorable default

experience.

Exhibit 182: P/C company insolvency rates (1981 ~ 2002)

0.30.27

0.37

0.9

1.8

1

0.83

1.4

1.8

1.27

1.44

2.05

1.2

0.58

0.210.28

0.79

0.6

0.23

1.021.03

1.33

0

0.5

1

1.5

2

2.51981

1982

1983

1984

1985

1986

1987

1988

1989

1990

1991

1992

1993

1994

1995

1996

1997

1998

1999

2000

2001

2002

Rate

of

Inso

lven

cy

Source: A. M. Best

Exhibit 183: ‘C’ rated insurers experienced the highest insolvency/impairment

rates three years later

0.02

0.57

0.86

1.45

1.72

2.61

0.93

1.25

0

0.5

1

1.5

2

2.5

3

A++/A+ A/A- B++/B+ B/B- C++/C+ C/C- D/E NR/NF

Ra

te o

f In

so

lve

nc

y

Source: A. M. Best

31 March 2006


P&C failures – a historical prospective

Over the past 10 years, the number of insolvencies has peaks and troughs. After reaching

a peak in 1992, when Hurricane Andrew hit local Florida property insurers, the failure rate

decreased and has remained at much lower levels for years.

In the late 1990s, reinsurers were under significant pressure from shareholders to expand

market share to generate significant top-line growth. Due to a limited number of viable

acquisition candidates in the reinsurance market, they began to offer cheap reinsurance

protection to stimulate growth. Some weaker primary insurers benefited from this, taking

advantage of lower reinsurance costs, and were able to remain in the market by passing

on substantial losses to reinsurers. Subsequently, reinsurers raised their rates, leading to

a series of insolvencies among insurers. The late 1990s was analogous to the mid 1980s,

when primary carriers engaged in cash flow underwriting (a strategy which justifies price-

cutting when the additional cash flow from increased market share provides investment

income which offsets higher underwriting losses), which led to insolvencies.

Over the past 10 years, insolvencies were predominantly driven by deficient loss reserves

(51%), followed by rapid growth (10%), the strain of discontinued operations (9%),

catastrophic losses (8%), the impairment of an affiliate (8%) and allegations of fraud (5%).

While there were a number of P&C insolvencies (i.e., 18, 7, 30, 30 and 38 insolvencies for

1998 to 2002, respectively), insurance debt and trust preferred defaults are somewhat

harder to find, which may be partly due to the fact that not all insurers have issued debt,

preferreds or surplus notes. However, to provide some color about distressed

insurers/reinsurers and the reasons behind defaults, we describe one incident announced

in November 2003, and three recent and rather significant insolvencies that resulted in

debt defaults. Reasons for these failures include under-pricing, deficient reserves,

aggressive underwriting and over-expansion.

Examples of

defaults in trust

preferreds or

surplus notes

31 March 2006


2003

PMA Capital Corp (PMACC), the holding company, has two major arms, the PMA Re and the PMA Insurance Group

(PMAIG). PMA Re was established in 1970 to provide property, casualty and specialty reinsurance. The PMAIG was

established in 1915 to provide specialized insurance in worker's compensation and disability insurance products and

services in the eastern US.

PMA Capital Corp issued $32.5 million trust preferred securities in June 2003 and $57.5 million in senior notes to help

improve financial strength and flexibility, some of which were used to pay down the more restrictive credit facility. As of

November 2003, the company has $4.5 billion in assets and $617 million in equity, along with $186 million in long-term

debt.

In November, the company took a $150 million pre-tax reserve charge that stemmed from the losses in its troubled

reinsurance operations for the accident years 1997-2000. Following the reserve charge, PMACC has been looking into

various alternatives to restructure the reinsurance arm. Some of the options include run-off, which means no new policies

would be underwritten and existing policies would terminate through attrition, or to sell off renewal rights to another

insurance company. Currently, PMACC has reached tentative agreements with Imagine to sell renewal rights to PMA Re's

existing policyholders. The news of reserve charges also prompted lawsuit filings on behalf of the debt holders, accusing

the company of providing inaccurate income information and maintaining inadequate reserves.

Following the announcement and the suspension of common stock dividends, PMA's equity price dropped 62%. Moody’s

downgraded PMA Capital Corp’s senior unsecured debt to Ba3 from Ba1, and preferred stock to B2 from Ba2, while Fitch

downgraded the senior unsecured debt to B+ from BB+. Moody's also expressed concerns over PMACC's organizational

structure that may result in regulatory constraints for PMACC to receive dividend payments from PMAIG, which may

impair PMACC's ability to service it debt obligations, i.e., including likely reducing the chance for timely payments to the

trust preferred holders.

2003

Lumbermen’s Mutual Casualty Co. is in default on $700 mm of surplus notes. Lumbermen’s primarily became distressed

from its significant concentration in California workers' compensation business, a market that was severely under-priced

from 1996-2000. Lumbermen’s also fell victim to the aggressive tort environment for asbestos. In 2001 and 2002,

Lumbermen’s took very significant charges for workers' comp and asbestos. In April 2003, the California insurance

regulator ordered the company to stop paying on the surplus notes given that the company had a deficit to pay claims on its

written policies.

2003

A large reinsurer, Trenwick Group, defaulted on about $250 mm of senior debt and preferred securities. Trenwick had

about $76 million of debt, $68 million of trust preferred securities, $75 million of preferred stock and the balance in

convertible preferreds. Trenwick had very aggressively written business since 1995, and also did not properly price and

reserve for the policies. Consequently, it became distressed when it incurred a very significant reserve charge (unrealized

loss) in 2002.

31 March 2006


L&H companies

We summarize L&H insurers’ historical financial impairment experiences as follows:

• Exhibit 184 suggests that the percentage of financially impaired companies (FICs)

dropped considerably throughout the 1990s, after peaking in 1991, as a strong

economy, low interest rates, and robust stock and real estate markets helped to

strengthen insurers.

• The average annual insolvency rate for L&H companies was 0.48% (with a standard

deviation of 0.45%) across the 26 years (Exhibit 184). Roughly, this implies a 4.8%

(=0.48%*10) 10-year default rate, assuming a constant annual default rate of

0.48%, which implies a low Triple-B rating, according to Moody’s corporate default

rate.

• The number of financial-impaired insurers has averaged less than four per year

since 1995 (excluding 1999),187 and most were relatively small, less efficient health

insurers, which had inadequately priced their products or could not adapt to rapidly

changing market conditions. Key contributors of financial impairments include

inadequate pricing (24%), affiliate problems (22%) and rapid growth (16%) (Exhibit

185).

• Impairments by product line indicate health insurance-related companies suffered

the most (46%), followed by life insurers (39%) and annuity writers (15%). Difficult

competitive and regulatory conditions continue today for health insurers.

Consequently, there has been considerable consolidation among health carriers, as

those that could not compete effectively have either been acquired, chosen other

lines of business, or have shut their doors.

Exhibit 184: L&H financial impairment rates (1976~2001)

0 .0 0%

0 .14%

0 .28%

0 .07%

0.20 %0.19%0 .19%

0 .6 0%

0 .2 9%

0.17%

0 .2 1%

0 .68%

0.4 2%

0 .72%

0 .77%

1.8 6%

1.2 3%

0 .8 8%

0 .76%

0 .13%

0 .2 7%0 .29%

0.44 %

1.35%

0.31%

0 .0 7%

0.00%

0.20%

0.40%

0.60%

0.80%

1.00%

1.20%

1.40%

1.60%

1.80%

2.00%

1976

1977

1978

1979

1980

1981

1982

1983

1984

1985

1986

1987

1988

1989

1990

1991

1992

1993

1994

1995

1996

1997

1998

1999

2000

2001

% L

&H

Im

pair

me

nts

Source: AEGON IMD Structured Products Research, S&P, Fitch.

187

The impairments of 1999 were an anomaly. In spring of 1999, regulators were beginning to uncover the alleged theft of funds from the seven companies controlled by the fictitious Thunor Trust. According to A.M. Best, of the 21 documented FICs that year, eight were tied to the activities of Martin Frankel, who allegedly absconded with over $200mm of assets from these insurers and fled the country. These companies are either currently under regulatory supervision or are in the process of being liquidated. While alleged fraud has occurred from time to time, the size of the alleged theft and the incredible chain of events surrounding Martin Frankel were extraordinary.

31 March 2006


Exhibit 185: Primary causes for L&H financial impairments (1976~2000)

24%

22%

16%15%

10%

6%

4%3%

0%

5%

10%

15%

20%

25%

Inade

q pr

icing

Aff i

liate

pro

blem

s

Rapid g

row

th

Overs

tate

d ass

ets

Alleged

fra

ud

Signific

ant c

hg in b

usiness

Misc

Reinsu

rer fa

ilure

Source: A. M. Best, Credit Suisse

Regulation and Disclosure Regulation of the insurance industry is centered at the state level. There are about 50

state insurance departments (and one for the District of Columbia) that regulate industry

activities. State regulators serve three primary functions. First, they monitor the financial

condition and claims-paying ability of companies operating in their state. Second, they

serve as consumer “watchdogs,” ensuring that policyholders aren’t overcharged or

discriminated against. Finally, regulators try to ensure that essential risk coverage is

readily available.

The National Association of Insurance Commissioners (NAIC) coordinates the activities

among the individual states. The NAIC proposed a bill, which later became known as the

McCarran-Ferguson Act (enacted March 9, 1945), which laid the framework for insurance

regulation. It declared the following:

• It was the intent of Congress that state regulation of insurance should continue and

that no state law relating to insurance should be affected by any federal law unless

such law is directed specifically at the business of insurance;

• All states impose investment limitations;

• States regulate rates and expenses;

• States control agents’ activities; and,

• States have control over contractual provisions and their effects on the consumer.

All insurance companies must file annual statutory statements of their income and

financial condition in accordance with generally uniform statutes with the NAIC

(www.naic.org). These statements, also known as convention statements, report statutory

results. The accounting under which these results are compiled is termed statutory

accounting. All public insurance companies are required to file a 10K with the SEC.

Regulation of the

insurance industry

is done on a state-

by-state basis

The NAIC

coordinates the

activities among

states

All insurers must

file annual

statements

31 March 2006


Insurance sector outlook P&C companies: improvement in pricing and underwriting disciplines

Moody’s outlook for the US personal lines insurance industry is stable, as US P&C insurers

shift their focus towards restoring underwriting discipline, following a period of weak

profitability.188 Results in the personal P&C lines started to show improvement in 2002 from

the pricing and underwriting perspective, as the current market seems to be defined by

participants ’ adoption of a “rate increase and profit ” versus “growth ” model. Even those

insurers with business strategies that historically focused on market share growth have

begun to exhibit a renewed sense of pricing discipline in the major product lines.

These developments have not occurred in a vacuum. In fact, several favorable market

characteristics have disappeared since the late 1990’s, including redundant reserve

positions, benign automobile loss cost trends and high investment yields. Personal

insurers have responded to the soft market cycle with important structural advances,

namely fundamental improvements to their operational capabilities, in terms of both

technology and underwriting. Moody’s expects that these changes will not only help firms

achieve and sustain lower combined ratios,189 but may help to moderate the amplitude of

the industry ’s inescapable cyclicality. With better information available sooner, carriers

should be positioned to react more quickly to deteriorating pricing conditions that, if left

unchecked, can choke profitability.

P&C companies have been imposing stricter underwriting standards and charging higher

premiums. For example, the persistent bullish trend in auto premium rates as compared to

auto repair costs is apparent. According to the consumer price index, personal auto

insurance premiums rose at an annual rate of 7.9% in the 12 months ended July, and

remain at a relatively high rate. In comparison, auto maintenance and repair costs,

representing a little less than half of personal auto insurance costs, rose at a 3.3% annual

rate for the 12 months ended July and remain low. The difference between personal auto

insurance premiums and auto repair and maintenance costs is respectable, i.e., it stayed

positive for the 26th straight month, though it has declined a bit.

CREDIT SUISSE’s insurance analysts believe bond spreads in the P&C sector continue to

benefit from strong technicals and healthy operating results across most of the companies

due to the persistent “hard” market (as in not soft), a phase in the underwriting cycle where

premium rates are increasing.190 Second quarter 2003 earnings results showed good

operating strength from the major insurers as a result of this persistently “hard” market,

and most of the companies in the quarter beat consensus estimates.

However, our analysts caution that evidence continues to emerge that the “hard” market

may be softening in the form of companies’ elaborations on financial results, their stronger

emphasis on growth through higher policy counts rather than through higher premium

rates, and industry studies showing such trends. While premium rate increases are

slowing, the current increases continue to build upon a sizable base of increases over the

past three years, and current premium rates are contributing to favorable returns.

188

“US Property & Casualty Personal Lines Insurance Industry Outlook,” Moody’s, April 2003. 189

The combined ratio is the sum of loss ratio and expense ratio. If the combined ratio is less than 100%, the difference is the underwriting profit margin. If the combined ratio exceeds 100%, underwriting was unprofitable - there was an underwriting loss. 190

CSFB Insurance Monthly, August 2003, fixed income research.

An example - a

bullish trend in auto

premium rates

Caution - premium

increases are

slowing

31 March 2006


L&H companies: strong core credit strength and more expected consolidations

Moody’s believes that L&H companies continue to benefit from a number of core credit

strengths.191 These strengths include adequate capital supporting conservative balance

sheets, predictable and profitable blocks of seasoned liabilities, and tax-favored product

offerings. Credit profiles of US life insurers will continue to be driven by economic,

demographic, and competitive trends and the related impact of customer preferences for

various products and product delivery.

Moody’s also suggested the strategic rationale for continued consolidation remains strong,

despite the low level of such activity during 2002. A weak economy and slower revenue

growth have negatively affected the valuation of many companies, creating opportunities

for some larger, better capitalized companies to increase their scale, diversification, and

distribution resources through acquisitions. Near-term acquisitions of entire companies

may not be attractive for many companies, but Moody’s believes that block acquisitions

will eventually become more prevalent. Larger diversified companies may seek to shed

businesses that have limited scale and are not performing up to

shareholders ’expectations, or sell units that are not part of their core competencies to free

up capital for other operations. As a result, product-focused niche players should benefit

from consolidation, as business line acquisitions could improve their competitive position.

ITP CDOs’ structural enhancement and other tidbits For ITP CDO investors, it is also important to focus on: 1) structural enhancement; 2) cash

flow analysis; and, 3) the relative value comparison to other CDO products. We will

discuss these aspects in turn.

There are also some unique structural protection features common to these deals that are worth noting. These include:

Diversion of excess spread. For some deals, a preset portion of the excess spread that is available for income note distribution will be used to pay down principal of the most senior notes (typically in year 8~10) until all senior notes have been paid in full.

ITP CDOs have a long legal final (typically 30 years). 192 Two common structural enhancement features seen in long-maturity CDOs are debt turboing and auction call, both of which are intended to reduce the average life of CDO debt.

Using excess spread to pay down the most expensive liability first,193 usually the Triple-B rated tranche in an ITP CDO, can increase the amount of future excess cash flow and shorten the Triple-B’s average life. For tranches senior to the Triple-B, credit enhancement is not affected, because only excess interest (which would have otherwise been paid to equity) is applied to pay down the Triple-B, and this is often conditional on the satisfaction of senior and mezzanine coverage tests. In essence, subordination is “replaced” with OC (i.e., over-collateralization).

Auction call. Most deals have the ability to solicit auction bids for the entire portfolio of

securities whereby sale proceeds are used to pay off the notes. Typically beginning from

year 8~10, the trustee will solicit auction bids for the purchase of all the remaining

collateral. If the net proceeds from the highest bid are equal to or greater than the principal

amount of the senior notes and mezzanine notes (including accrued and unpaid interest

191

“Credit issues and trends for US life insurance,” special comment, May 2003, Moody’s. 192

In the generic deal that we illustrate later, we assume the collateral are called at a 5% annual rate after year five, the end of the non-call period, with 75% of the collateral being called at year ten, the first auction call date. This leads to an average life of 9.25 years for the collateral under a zero default assumption. The average life may extend if the collateral call rate declines. 193

This is often conditional on the satisfaction of preset equity return targets.

Likely continued

consolidation

Diversion of excess

spread

Shortening the

average life of CDO

debt

Turbo

Action call

31 March 2006


and fees/expenses), the trustee will sell all the collateral. Sale proceeds will then be used

to redeem the senior notes and mezzanine notes on the payment date immediately

following the auction date with any additional amount going to the income noteholders.

Barring any unexpected credit deterioration in the pool, it is likely that the auction call can

be exercised. At the auction call date, two occurrences are likely to have happened: 1) the

collateral would have seasoned and shortened its remaining average life, possibly being

sold at tighter spreads (and/or higher prices); 2) the Triple-B would have been partly

paid down from debt turboing, reducing the amount of outstanding CDO liabilities.

Both of these may result in an in-the-money auction call; i.e., the value of the

collateral pool being greater than the value of the liabilities. Separately, after a CDO’s

regular non-call period expires, equity holders are increasingly likely to call the deal,

as the CDO should have de-levered from Triple-B turboing and collateral prepayment,

which reduces the leverage and arbitrage (and the deal’s seasoned and shorter

collateral could indeed be recycled into a new CDO). We caution that should collateral

credit deteriorate, both the auction call and the regular call become less likely to be in-

the-money.

Additional principal pay down. Similar to bank trust preferred CDOs, this allows payments

to senior notes by the amount of defaulted or deferring assets even if senior OC tests

remain in compliance. Payment for the additional principal paydown is made by using

excess spread that otherwise would have been distributed to the income notes (equity).

A key part of CDO cash flow analysis is examining the internal rate of return (IRR) profiles.

As a generic example, we look at a representative ITP CDO with a $300 mm capitalization,

with 62% in ‘AAA’, 12% in ‘AA’, 10% in ‘A’, 6% in ‘BBB’ and 9% in equity. Exhibit 186

shows the internal rate return profile (IRR) for the CDO debt. Please see the footnotes of

Exhibit 186 for modeling assumptions. Exhibit 186 indicates that the Triple-A, Double-A,

Single-A and Triple-B can sustain about 9.3%, 7.4%, 4.5% and 3.5% constant annual

collateral default rates (CDR), respectively, before each bond begins to lose yield. These

breakeven rates suggest that the CDO debt has a very reasonable amount of protection

against the historical default rates, 0.94% and 0.48% annually for P&C and L&H

companies, respectively, and averaging about 0.80% for a typical 70/30 P&C and L&H

blended ITP CDO collateral pool.

Exhibit 186: IRRs for a generic ITP CDO *

-8%

-6%

-4%

-2%

1%

3%

5%

7%

9%

11%

0% 1% 2% 3% 4% 5% 6% 7% 8% 9% 10%

CDR

IRR

AAA AA

A BBB

* We assume the collateral are called at a 5% annual rate after year five, the end of the non-call period, with 75% collateral being called at year

ten, the first auction call date; no deferral of interest payment on the collateral; constant default starting immediately; 10% recovery with no lag;

turbo ‘BBB’ with a 23% equity cap. Spread assumptions: collateral L+385bp, liability ‘AAA’ L+110bp, ‘AA’ L+140bp, ‘A’ L+205, ‘BBB’ L+375bp.


Additional principal

pay down

IRR profile

31 March 2006


There are also attractive relative value opportunities, as ITP CDO notes offer a substantial

spread pick-up over other more established CDO products. While bank trust preferred

CDOs Triple-As reached 77 bps on average this year and more established CDO products

such as HY CLOs and SF CDOs have typically priced below 60 bps, new issue ITP CDO

Triple-As generally offer an above 100 bps spread. This also comes with a greater credit

enhancement. For example, the subordination to Triple-As is on average 36%~44%,

averaging 39% across the five outstanding ITP CDOs (which converts to 156%~179%

OC),194 comparing to an average of 26% for a HY CLO.

Closing The application of the CDO technology to insurance risk creates a “win-win” in that it also

provides an opportunity for mainstream fixed income investors to buy pooled insurance

trust preferred risk at an attractive spread. In general, insurance deals have priced wider

than bank trust preferred deals, partly attributable to a new product premium (including

less investor familiarity with the collateral) and the perception of higher risk in state-

regulated small insurers vs. federally regulated banks. ITP CDOs also offer a higher

yielding investment opportunity, although one that comes with a give-up in liquidity.

For ITP CDO investors, collateral due diligence is important. Some deals employ a

manager to select/originate the initial collateral. To the extent that the manager has

extensive expertise in the insurance industry, this can offer additional comfort to investors.

We believe relatively well-capitalized smaller insurers with strong underwriting discipline

will continue to outperform. As well, pooling a group of these small insurers (many with a

niche product focus) should benefit CDO investors by creating product line and

geographical diversification in the pool. In addition, some of these smaller insurers have a

higher likelihood of being acquired, potentially enhancing ITP CDO performance.

194

The upfront cushion between actual OC and the triggers is also clear, as the senior OC trigger is typically set between 125%~128% for Triple-As, indicating a cushion greater than 30%. A larger cushion reduces the likelihood of early amortization.

Relative value:

Spread pickup with

greater credit

enhancement

31 March 2006


An Introduction to REIT Trust Preferred CDOs195

Trust preferred (TruPS) CDOs have grown from a niche market to a market mainstay

since their inception in 2000. As CDOs are driven by innovation, it’s no surprise that

TruPS CDOs have also evolved. The latest advancement in the TruPS CDO space is the

inclusion of REIT-issued trust preferred securities in CDO portfolios. While some CDO

pools have traditionally reserved small buckets for REIT TruPS, this year for the first time,

the majority of a CDO’s portfolio was comprised of REIT TruPS. Three REIT TruPS CDOs

totaling $2.6 billion have been issued so far (please see Appendix A), and we expect this

asset class to continue to gain momentum and investor interest.

In this section, we provide an introduction to REITs and REIT TruPS CDOs. As only a few

REIT TruPS CDOs have been priced so far, our focus is on defining REITs and their

historical performance and suitability as CDO collateral. We discuss the following:

• Define REITs – what are they from an equity and debt perspective;

• Discuss trust preferred securities and why they are a suitable financing platform for REITs;

• Outline the diversification benefits of REITs;

• Discuss key credit considerations in evaluating REITs;

• Review the credit performance of REITs;

• Provide a REIT Sector Outlook; and

• Discuss the REIT TruPS CDO platform.

REITs Defined What is a REIT?

196

A Real Estate Investment Trust (REIT) is a tax-efficient pass-through entity that functions like a mutual fund for real estate investments. REITs own, and in most cases, operate income-producing real estate. Additionally, some REITs also engage in real estate financing. REITs were created to provide smaller investors access to large-scale, income-producing real estate, with the benefit of diversification through a portfolio of real estate assets managed by experienced real estate professionals. REITs are exempt from corporate taxation by way of The Real Estate Investment Trust Act, subject to certain statutory requirements. These requirements include:197

• At least 90% of taxable income must be distributed in common and/or preferred stock dividends each year;

• At least 75% of the book value of total assets is invested in real estate equity and/or mortgages;

• At least 75% of gross revenue is from rents and/or interest on mortgages; and

• Not more than 50% of the REIT is owned by five or fewer individuals. There are essentially three types of REITs: equity, mortgage and hybrid.

• Equity. Equity REITs own and operate income-producing real estate. Their

activities may include leasing, development, and tenant services. Equity REITs

must acquire and develop properties primarily to operate them as part of its own

portfolio, and not to resell them once developed.

195

This section was originally published in "The CDO Strategist", Issue #8, September 30, 2005. 196

This section makes extensive references to data and materials found in: "Frequently Asked Questions About REITs", The National Association of Real Estate Investment Trusts 197

"Real Estate Investment Trusts", presentation from CSFB REIT Debt Research, October 2004

31 March 2006


• Mortgage. Mortgage REITs lend money directly to owners/operators of real

estate or extend credit indirectly through the acquisition of loans or mortgage-

backed securities (MBS). Some mortgage REITs may have their own loan

servicing operations.

• Hybrid. Hybrid REITs own properties and originate loans to real estate

owners/originators.

Equity REITs account for the majority of publicly traded REITs, followed by mortgage and

hybrid types (see Exhibit 187).

As of September 2005, the National Association of Real Estate Investment Trusts

(NAREIT) reports approximately 200 REITs registered with the Securities and Exchange

Commission (SEC) in the United States that trade on one of the major stock exchanges,

with total assets exceeding $400 billion. Additionally, approximately 800 REITs are not

registered with the SEC and are not traded on a stock exchange. Exhibit 187 and Exhibit

188 show the breakdown of registered REITs by type and by market capitalization.

REIT Debt 198

A REIT, like any other corporation, issues a combination of equity and debt to finance its

operations, acquisitions, and long/short term funding needs. Over $12 billion in REIT

unsecured debt has been issued this year so far (as of 9/23/2005), well on track to another

active year in the primary REIT bond market (Exhibit 189).

Exhibit 189: Historical REIT Corp. Bond Issuance – Another Strong Year in 2005

His torical REIT Unsecured Debt Issuance ($ m m )

$1,680$2,140

$3,459

$4,426

$9,240

$13,786

$7,951$8,583

$9,570

$10,733$10,157

$ 16,250

$ 12,353

$0

$2,000

$4,000

$6,000

$8,000

$10,000

$12,000

$14,000

$16,000

$18,000

1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005

Source: Credit Suisse. As of 9/23/2005.

198

This section makes extensive references to a presentation from CSFB's REIT Debt Research: "Real Estate Investment Trusts", October 2004

Exhibit 187: Registered REITs by Type Exhibit 188: Registered REITs by Market Cap Sizes

Equity

90%

Mortgage

8%Hybrid

2%

> $2 bn

24%

$100 mm -

$2 bn

68%

< $100

mm

8%

Source: Credit Suisse, NAREIT. As of 9/1/2005. Source: Credit Suisse, NAREIT. As of 9/1/2005

31 March 2006


REIT debt has traditionally been favored by (predominately) buy-and-hold investors

because of two qualities differentiating them from corporate bonds. They include:

1) Strict, protective bond covenants; and

2) Relatively stable and predictable cash flows.

Unlike a typical corporate bond, REIT bonds have strict covenants designed to keep

overall leverage to a safe minimum, enforce financial discipline, and ensure that some

portion of the assets are unencumbered, which protects the bondholder from being fully

subordinated to mortgage lenders. A typical bond covenant package includes:

1) Total Debt/Total Assets < 60%;

2) Total Secured Debt/Total Assets < 40%;

3) EBITDA/Interest Expense > 1.5%;

4) Unencumbered Assets > 150% of Unsecured Debt.

The first three covenants are on an incurrence basis, which means additional debt cannot

be incurred if any of these covenants would be violated. By contrast, the unencumbered

asset test is on a maintenance basis, which means the covenants must be met at all times,

not just on an incurrence basis. Combined, these covenants help maintain the bond’s

rating stability. Moreover, the covenants, along with SEC oversight of REIT equity,

enhance the market transparency of REIT investments.

Because of the nature of a REIT’s assets, CREDIT SUISSE REIT Analysts view REIT

cash flows as more predictable compared to other corporate bonds. For example, sales

revenue is a primary source of cash flow for some corporate issuers and this source may

be highly volatile or subject to seasonal fluctuations in some industries.199 By contrast,

cash flow from real estate assets is contractual and generated on a property by property

basis. In addition, other sources of cash flow include:

• Advisory and management fee income;

• Retained cash flow from operations;

• Sale of assets;

• Equity issuance;

• Mortgage financing;

• Private capital; and

• Use of bank lines.

Investors should note however, that market shocks and company-specific difficulties could

impact cash flow negatively. Furthermore, occupancy rates, lease expirations, tenant

credit quality, prospects for rental growth, and the overall quality of a REIT’s portfolio all

contribute to the health of anticipated cash flows. Please see the section, REIT Evaluation,

for key considerations in evaluating REITs.

199

"Bonds - REIT Bonds", NAREIT Features, September/October 2004

31 March 2006


REIT Trust Preferred What Are Trust Preferred Securities and How Do They Work?

200

Since their inclusion as Tier 1 capital in 1996, trust preferred securities (TruPS) have been a popular financing mechanism for banks, utility companies, REITs, the insurance sector and more. TruPS are hybrid securities comprised of preferred equity issued by a special purpose trust, and debt issued by the company. For an overview of the trust preferred issuance structure, please see Appendix B. A TruPS promises to make periodic coupon payments and has a stated maturity (debt like), generally 30 years. Unlike debt, it is required to make the coupon payment only when the issuer is financially able (equity like). Otherwise, interest may be deferred for up to five-years, and the deferred interest is paid back on a cumulative basis. A TruPS is a bullet bond (not amortizing) with a 5- or 10-year non-call period. After this period, it is callable, typically at par (but not always). The “equity like” nature enables a TruPS to be qualified as equity for regulatory capital purposes. Its “debt like” attribute enables the coupon payment on the security to be tax-deductible for issuers, unlike other forms of equity. TruPS have been created by companies for their favorable accounting treatments and flexibility. Specifically, these securities are taxed like debt obligations by the IRS while maintaining the appearance of equities in a company's accounting statements in accordance with GAAP procedures. Upon liquidation of the issuing company, TruPS rank senior to the company’s preferred and common stock and junior to the company’s debt.

Applying Trust Preferred to REITs

For small to medium sized (SMS) REITs (< $2 bn market cap – see Exhibit 188), which account for the lion’s share of registered REITs, accessing the unsecured capital markets may prove difficult and costly. Often, the capital requirements of SMS REITs do not meet the market minimums for new-issue securities and the cost associated with underwriting and marketing equity can be expensive. Furthermore, since REITs must maintain certain leverage ratios, borrowing from the unsecured debt market may not be an option. Trust preferred financing is a natural fit into a REIT’s funding alternatives. Just like pooled bank and insurance trust preferred, REIT TruPS offer a more level playing field for smaller REITs (vs. larger ones), as the TruPS platform allows smaller REITs to achieve lower financing costs and faster execution. Exhibit 190 shows available aggregate REIT trust preferred issuance data for the past three years. There are several interesting take-aways from the table:

• While the legal maturity of REIT TruPS is 30-years, it is likely that the REIT issuer will call the TruPS and refinance at the end of the non-call period, typically 5 or 10 years. Because of this feature, spread levels over both the 10-year and 30-year UST are provided.

• REIT TruPS offer very attractive dividend and spread levels. This is partially due to the liquidity premium from the relatively small size of the REITs. Focusing only on rating and spread, even at a Single-B rating, which is many notches lower than the weighted average rating (WAR) of REIT TruPS, leveraged loans currently offer around 250 bps of spread compared to 324 bps on average for BB-/BB REIT TruPS.201

• 26 REITs were repeat issuers in the TruPS space, including one REIT that issued nine times over the three year period.

• Nearly half the REIT TruPS in each vintage were unrated.

200

"Diversified Bank Trust Preferred CDOs", CSFB CDO Research, October 2003 201

Based on CSFB's Leveraged Loan Index

31 March 2006


While REIT TruPS do not typically have explicit covenants protecting holders, they do

benefit indirectly from the implied protective covenants of REIT senior unsecured debt

mentioned in the previous section, assuming the REIT is also an issuer of debt. This

implied protection should support stable ratings and sufficient cash flow, barring any

industry or company shocks.

Diversification Benefits – A Look at Total REITurns One of the benefits of REITs is their ability to provide portfolio diversification due to their

low return correlation with returns of other assets. Not only are REITs diversified by type

and geographic concentration, but also by industry distribution. Exhibit 191 shows the

industry distribution among registered REITs.

Exhibit 191: Industry Distribution of Registered REITs

Office

17%

Retail

25%

Diversif ied

8%

Industrial

6%

Lodging/Resort

5%

Manufactured

Homes

1%

Health Care

5%

Self Storage

4%

Specialty

4%

Home Mortgage

Financing

6%

Apartments

14%

Commercial

Mortgage

Financing

2%

Mixed

Off ice/Industrial

3%

Source: Credit Suisse, NAREIT. As of 9/1/2005.

Performance among REIT industries varies considerably as not all real estate act alike.

Exhibit 192 shows the monthly total return correlations among constituent industries in

SNL Financial’s REIT Index from 1990 to August 2005. It is evident that certain industries

are much less correlated than others. Excluding Diversified REITs, Office and Hotel

REITs displayed the most negative return correlation at –0.059 while Multifamily and

Residential REITs showed the highest return correlation at 0.993. It’s also worthy to note

that Diversified REITs, which are typically comprised of multi-industry portfolios, were

negatively correlated with nearly every other industry.

Exhibit 190: Aggregate REIT Trust Preferred Issuance Data by Vintage since 2003

Vintage

Total REIT TruPS

Issuance ($mm)

Total #

Issued

Avg.

Size

Min

Dividend

Max

Dividend

Weighted

Avg.

Dividend

Avg.

Spread vs

10yr UST

Avg.

Spread vs

30yr UST

%

Rated WAR Largest Sector

2003 $5,256 56 $94 6.450% 11.000% 7.742% 375 283 53.6% BB+/BBB- Retail (29%)

2004 $5,871 61 $96 6.125% 9.750% 7.483% 339 260 54.1% BB/BB+ Diversified (18%)

YTD 2005 $2,463 27 $91 6.180% 9.125% 7.459% 324 294 51.9% BB-/BB Diversified (28%)

Source: Credit Suisse, SDC. As of 9/23/2005. Note that: WAR is calculated based on the S&P/Fitch rating for rated REIT TruPS.

31 March 2006


Exhibit 192: Monthly Total Return Correlations of REIT Industries (1990-2005)

Healthcare

Hote

l

Industr

ial

Div

ers

ifie

d/ O

ther

Offic

e

Reta

il

Resid

ential

Multifam

ily

Manuf H

om

es

Hotel 0.363

Industrial 0.362 0.530

Diversified/Other 0.545 0.445 0.533

Office - 0.024 0.057 0.125 - 0.102

Retail - 0.015 - 0.059 0.080 - 0.089 0.556

Residential 0.046 0.008 0.104 - 0.005 0.590 0.798

Multifamily 0.024 0.007 0.090 - 0.011 0.564 0.784 0.993

Manuf Homes 0.015 0.046 - 0.001 - 0.039 0.285 0.482 0.498 0.473

Self-storage 0.019 0.039 - 0.005 - 0.072 0.495 0.652 0.610 0.605 0.431

Source: Credit Suisse, SNL Financial. Figures reflect SNL’s REIT Index.

The diversification available within REITs translates into relatively low return correlation

with other asset classes. Exhibit 193 shows the monthly total return correlations between

several major market indices and the total returns of the NAREIT index and the CREDIT

SUISSE Liquid US Corporate Index (LUCI) for REITs, separately.202 Examining the LUCI

REIT bond index first, its correlation with other fixed income indices is very high while

correlation with equity and high yield indices appears relatively low. Note that the LUCI

REIT data only dates back to 2000/2001.

The NAREIT index, which provides a more compelling story, has a much longer history.

Over the 13-year time period between 1992 and 2005, return correlation between NAREIT

and the major market indices was relatively low across nearly all sectors. According to

Exhibit 193, the total return correlation between NAREIT and other market sectors was

among the lowest in each sector examined. The lowest correlations, 0.061 and 0.107,

occurred with the Merrill Lynch ABS and Mortgage indices, respectively, while the highest

correlation, 0.389 and 0.336, occurred with the CREDIT SUISSE High Yield and

JPMorgan Emerging Markets indices.

202

For more information on CSFB's LUCI bond index, please refer to "Introducing the Liquid U.S. Corporate Index (LUCI)", CSFB Index Research, November 15, 2002

31 March 2006


Exhibit 193: Monthly Total Return Correlations Among Other Market Sectors

ML M

ort

gage

ML A

BS

ML C

orp

LB

Aggre

gate

Bond

S&

P 5

00

DJ W

ilshire 5

000

JP

M E

mg.

Mark

ets

CR

ED

IT S

UIS

SE

Conv S

ecs

CR

ED

IT S

UIS

SE

HY

Index

CR

ED

IT S

UIS

SE

Lev. Loan

Index

ML ABS 0.873

ML Corp 0.843 0.834

LB Aggregate Bond 0.929 0.910 0.957

S&P 500 0.077 -0.066 0.158 0.053

DJ Wilshire 5000 0.045 -0.100 0.139 0.023 0.979

JPM Emerging Markets 0.407 0.305 0.478 0.403 0.513 0.519

CREDIT SUISSE Conv Securities 0.043 -0.073 0.155 0.026 0.751 0.846 0.472

CREDIT SUISSE High Yield Index 0.190 0.055 0.362 0.193 0.489 0.534 0.516 0.604

CREDIT SUISSE Leveraged Loan Index -0.100 -0.138 0.071 -0.054 0.137 0.168 0.046 0.235 0.486

NAREIT* 0.107 0.061 0.238 0.153 0.291 0.328 0.336 0.278 0.389 0.258

CREDIT SUISSE LUCI – REIT** 0.807 0.801 0.859 0.894 -0.306 -0.292 0.279 -0.138 0.133 -0.007

*NAREIT & all other market correlations reflect period from 1992-2005;

**CREDIT SUISSE LUCI REIT correlations reflect period from 2001-2005.

Source: Credit Suisse, CREDIT SUISSE Leveraged Finance Strategy, NAREIT.

We also compare the total return

correlations among bank, insurance,

and REIT equity (Exhibit 194). Again,

REITs show relatively low equity return

correlation with bank and insurance,

based on the SNL and NAREIT indices

between 1992-2005. While we note that

this is not the same as default

correlation, REITs may provide

diversification benefits when pooled with

bank and insurance securities.

REIT Evaluation Key Credit Considerations

In general, the evaluation of REIT credit should include a thorough analysis of the

company’s fundamentals. CREDIT SUISSE’s REIT Debt Analysts have selected a

number of key credit considerations worth assessing.203 These include:

• Management – reviewing the creditability, operating history, strategies, and

possible succession issues of the managers of real estate held by the REIT. The

REIT’s length of time as a public company is also worth considering.

• Ownership/Corporate Structure – reviewing the REIT’s structural features,

which may impact credit quality: traditional REIT versus UPREIT or downREIT,

and any joint venture agreements.204

203

"Real Estate Investment Trusts", presentation from CSFB REIT Debt Research, October 2004 204

An UPREIT is a structure in which the REIT does not own a direct interest in properties, but rather in an umbrella partnership that owns interests in properties. For this reason, this umbrella partnership is generally referred to as the operating partnership. A side benefit of the UPREIT structure is that operating partnership units can be used as currency to acquire properties from owners who would like to defer taxes that would come due if the property(ies) were sold or swapped for stock. In response to this advantage of the UPREIT structure, a number of non-UPREITs have created so-called downREITs. This makes it possible for them to buy properties using downREIT partnership units. The effect is the same, however the downREIT is subordinate to the REIT itself, hence the name. (Source: Realty Stock Review)

Exhibit 194: Return Correlation Among

Bank, Insurance and REITs (1992-2005)

SN

L

Bank &

Thrifts

SN

L Insura

nce

SNL Insurance 0.7834

NAREIT 0.4820 0.4276

Source: Credit Suisse, SNL Financial. As of 9/23/2005.

31 March 2006


• Asset/Property Profile – assessing the property types, portfolio age and quality,

geographic distribution and tenant quality relative to the markets and locations

that the REIT and its competitors operate in.

• Financial Flexibility – assessing the REIT’s ability to provide short/long-term

funding; the refinancing risk of near-term obligations, secured debt (long dated

and free-and-clear properties will enhance flexibility), and debt covenants, which

could be too restrictive.

• Liquidity – assessing the cash flow and available borrowings under the bank

credit facility relative to near term obligations, including debt maturities and capital

expenditures.

• Capital Structure/Leverage – assessing the level of leverage; leverage varies by

property type, but REITs generally strive to maintain debt levels below a

percentage of total market capitalization, often between 40% and 50%.

• Cashflow Considerations – unlike most corporations, REITs are required to

distribute 90% of their taxable income. Therefore, common and preferred

dividends should be considered a fixed charge.

Besides these credit considerations, it’s also worthwhile to evaluate a few key equity ratios

to gauge the REIT’s operating performance. Specifically, the industry standard

methodology for evaluating a REIT's earnings potential is to review the Funds from

Operations (FFO). FFO excludes the following from the net income figure: depreciation

and amortization costs; gains and losses from extraordinary items; gains or losses from

debt restructuring; and, gains or losses from sales of real estate. In this way, the FFO

provide a more accurate assessment of real estate value versus the standard GAAP.

Credit Performance From a ratings and default standpoint, REITs have faired well. While REITs have been in

the market since the 1960’s, it was not until the early 1990’s that they experienced significant

growth and maturation into today’s REITs (Exhibit 195). For this reason, it makes sense to

focus on the last decade or so in evaluating the credit performance of REIT debt.

Exhibit 195: REIT Growth Picked Up Steam in the Early-Mid 90’s

$0

$50,000

$100,000

$150,000

$200,000

$250,000

$300,000

$350,000

1981

1982

1983

1984

1985

1986

1987

1988

1989

1990

1991

1992

1993

1994

1995

1996

1997

1998

1999

2000

2001

2002

2003

2004

Ma

rke

t C

ap

ita

liza

tio

n (

$m

m)

Hybrid

Mortgage

Equity

Source: Credit Suisse, NAREIT

31 March 2006


Standard & Poor’s published its first rating transition study for U.S. REITs in June 2005,

covering the 11-year period from 1994-2004.205 Exhibit 196 shows the average 5-year

cohort transition matrix for REIT debt:

Exhibit 196: Average 5-Year Cohort REIT Transition Matrix (1994-2004)

From/To A BBB BB B CCC/C D N.R.

A (%) 74.1 20.7 0.0 0.0 0.0 0.0 5.2

BBB (%) 1.7 64.5 5.1 1.7 0.0 0.0 27.0

BB (%) 0.0 22.8 40.4 5.3 0.0 0.0 31.6

B (%) 0.0 0.0 100.0 0.0 0.0 0.0 0.0

CCC/C (%) NA NA NA NA NA NA NA


The matrix reveals several interesting observations:

• There were no downgrades to CCC or below and no issuer defaults during the 11-

year study period.

• Of the 41 REIT ratings that were withdrawn (set to N.R. – not rated) by S&P

during the study period, 76% were because of mergers & acquisitions among

REIT issuers, reflecting the consolidation wave of the late 90’s

• REIT ratings are clustered in the BBB category; nearly 70% of all original ratings

were in the BBB bucket. While only 64.5% remained at BBB (largely attributed to

the 27% withdrawn BBB ratings), only 6.8% of BBB ratings were actually

downgraded on average during a 5-year period.

• There were little or no REITs in the best and worst rating categories. Of the few

that were initially rated B, all were upgraded to BB. 22.8% of REITs originally

rated BB were upgraded to investment grade.

While the results reflect a subset of the REIT universe (those actually rated), the rating

transitions study does point to favorable results for REIT credit since its expansion in the

early 90’s. Most promising is the absence of defaults and the ratings stability of the asset

class. Exhibit 197 compares the 5-year cohort ratings stability of REITs with that of

corporate bonds and CMBS, after removing withdrawn ratings. REIT ratings stability at

the BBB level, which accounts for the majority of rated-REITs, is significantly higher than

corporates and just slightly below CMBS. Interestingly, REIT ratings stability actually

increases as you go down in credit, but this may be due to the small sample size.

Exhibit 197: Comparative Stability Ratios (1994-2004)*

Stable/Upgrade Ratio REITs (%) Corporate (%) CMBS (%)

AAA NA 82.9 98.5

AA NA 62.4 97.2

A 76.9 77.3 94.5

BBB 89.2 80.9 91.4

BB 94.1 64.3 77.2

B 100.0 53.2 74.2

CCC/C NA 20.0 76.2

*Based on Average 5-Year Cohort REIT Transition Matrix, N.R. removed


205

"Rating Transitions: A Generally Favorable First Decade for U.S. REIT Ratings", S&P, June 8, 2005

31 March 2006


REIT Outlook REITs have performed well over the last five years. REIT credit spreads are trading at

historically tight levels and the trend is expected to continue. CREDIT SUISSE’s REIT

Debt Analysts currently recommend the sector as “Overweight” on the heels of strong

sector fundamentals and technicals, stable economic growth, and limited supply in REIT

paper.206 As rates remain low and economic growth remains steady, CREDIT SUISSE

Analysts believe that REIT bonds at current levels have value. Although investment-grade

spreads widened earlier in the year, the Analysts view this as a result of interest rate

concerns, inflation fears, and corporate headline news, rather than weakness in REIT

fundamentals or technicals. Credit ratings are expected to remain stable or improving

throughout the remainder of the year.

For specific REIT sectors, CREDIT SUISSE Analysts observe declining vacancy rates and

growing rents in a few property markets as signs of recovery in real estate fundamentals.

Specifically:

• Industrial and Office sectors appear to have bottomed out. Office vacancies

should continue to decline in 2005 but it may take another year or more to see

landlords regain pricing power. Demand in Industrial REITs appears to be

strengthening but landlords have no pricing power; speculative development still

hurts near-term prospects.

• Multifamily appears headed toward recovery, although supply/demand

fundamentals are still out of balance in many regions. Occupancy rates show

improvement in 2005, although near-term supply should continue to exceed

demand as new construction continues unabated.

• Retail outlook is robust as retailers are expanding, bankruptcies and store

closings are running below 2004 levels, and a lack of new mall construction and

consolidation have changed leasing dynamics.

The rating agencies share similar views as CREDIT SUISSE REIT Debt Analysts. As

mentioned, S&P recently published its rating transitions for REITs, reflecting the first

decade of performance.207 S&P views REITs as stable to positive, reflecting the generally

favorable ratings stability and absence of defaults over the last 11 years. Also reflecting

the favorable default history, Fitch has a stable to positive outlook on REITs as well,

particularly for mortgage/hybrid and lodging REITs. According to Fitch, REIT performance

has and will continue to benefit from macroeconomic factors, such as job growth, and

improving fundamentals. Moody’s has a slightly less positive outlook on REITs, but still

stable. However, on September 28, 2005, Moody’s upgraded Simon Property Group, a

retail REIT, to Baa1 citing strong performance in the retail sector and sound

fundamentals.208 This may be a signal for a more positive outlook from Moody’s.

Pooling it All Together: REIT TruPS CDOs The application of CDO technology with REIT TruPS creates a “win-win” for both SMS

REITs and investors in that it provides an opportunity for mainstream fixed income

investors to buy pooled REIT risk with industry and geographic diversity at an attractive

spread while providing REIT issuers, particularly those with non-investment grade ratings,

relatively cheaper and more efficient access to the capital markets. While the number of

issued REIT TruPS CDOs has been limited so far, there are some common characteristics

of these deals, which we discuss briefly.

206

"REITs: 2005 Outlook", presentation from CSFB REIT Debt Research, August 2005 207

"Rating Transitions: A Generally Favorable First Decade for U.S. REIT Ratings", S&P, June 8, 2005 208

"Moody's Upgrades Simon Property Group's Senior Debt to Baa1; Stable Outlook", Moody's, September 28, 2005

31 March 2006


Collateral Composition and CDO Structure

Exhibit 198 shows some basic details of the REIT TruPS CDOs issued so far:

For the most part, the majority of REIT TruPS collateral in CDO pools have been issued by

registered REITs. In addition, approximately 50% of the pools pay floating-rate, 13% pay

fixed-rate, and the remainder pays hybrid, whereby the coupon is fixed for the first 5-10

years and switches to floating thereafter.

The structural features of the deals are similar to other CDOs in the market. The cash flow

waterfall pays sequentially with an equity cap currently around 18%: excess spread is

used to turbo the BBB tranches. Like other CDOs, cash flow is diverted from junior

tranches to senior tranches if overcollateralization tests fail.

A few variations from typical CDOs include:209

• OC haircuts for REITs that fail to meet two out of the following three financial

performance tests: 1) interest coverage tests; 2) total debt to total capitalization

tests; or 3) tangible net worth tests.

• Additional OC haircuts for REITs that violate two out of the following three tests:

1) if the REIT eliminates the common dividend payout or does not pay for two

consecutive quarters; 2) if the REIT is unable to maintain a certain fixed-charge

coverage ratio; or 3) if any monetary covenant is not cured for 30 days.

REIT TruPS CDOs currently price with a new product premium. Compared to recently

issued hybrid bank & insurance TruPS CDOs, REIT TruPS CDOs offer a 5 to 45 bps

spread pickup and offer higher credit enhancement levels across the capital structure.

209

Fitch Presale Reports.

Exhibit 198: Basic Details of REIT TruPS CDOs issued so far

Deal/Structural Information

Average Deal Size ($mm): $850.67 Management Type: Static Pool

AAA Credit Enhancement (%): 42.4% Auction Call: 10 Years

AA Credit Enhancement (%): 28.6% Non-Call Period: 5 Years

A Credit Enhancement (%): 18.0% Maturity: 30 Years

BBB Credit Enhancement (%): 8.8%

BB+ Credit Enhancement (%): 7.8%

Average Equity Size: 8.8%

Collateral Information

Pool Concentration

REIT/REOC TruPS * 87% WAS: 306 bps

REIT/REOC Snr/Sub Notes 10% WAC: 5.13%

CMBS 3% Max Issuer Concentration: 3.75%

Source: Credit Suisse, Fitch Presale Reports, Bloomberg, MCM, IFR

* REOCs (Real Estate Operating Companies) are similar to REITs except they are taxed as ordinary corporations, but are not subject to the

same restrictions as REITs. REOCs do not need to distribute any dividends. This allows REOCs to have superior operating flexibility under

certain stressed circumstances compared to equity REITs. (Source: Fitch)

31 March 2006


A Note on Rating Agency Methodologies

The rating agencies have not yet formally published their rating methodologies for rating

REIT TruPS CDOs. We do not provide details for each methodology however, it is

important to review a few key differences across agency approaches.

Moody’s uses a more stringent approach by taking a pool wide perspective of the REIT

TruPS collateral to generate a pool wide rating and default probability across all unrated

REITs (which accounts for the majority of REIT collateral in the CDO) in addition to the

public Moody’s ratings where available, and assumes a 15% recovery rate across all REIT

TruPS.210

By contrast, Fitch and S&P will have their respective REIT sector specific analysts review

each trust preferred issuer and determine suitable ratings, default probabilities, and

recovery rates for each unrated issuer.

We expect rating agency treatment of REIT TruPS CDOs to become more refined as the

product develops and structures become more established.

What’s Next?

While REIT TruPS CDOs are still in the early stages of their development, we believe the

asset class is positioned for robust growth over the next year. REITs have a strong and

stable track record over the last decade. Their relatively low correlation within REIT

industries and across broader market sectors makes them suitable for CDOs. Given the

attractive asset level spreads, new product premiums, relatively higher equity returns, and

a positive REIT outlook, investors should consider REIT TruPS CDOs.

Because of the limited registered REIT universe (about 200 REITs), we expect more

hybrid pools to be issued. These hybrid pools may consist of other trust preferred

securities, including bank and insurance TruPS, and/or possibly other structured finance

credits. REITs should provide added diversification benefits to these assets classes.

Furthermore, CDO issuers may move into more private, non-registered REITs to source

trust preferred collateral.

210

"Trust Preferred Market Update and REIT CDOs", Moody's 5th Annual U.S. CDO Investor Briefing, September 7, 2005.

31 March 2006


Appendix A. REIT TruPS CDOs Priced as of September 2005

Taberna Preferred Funding I - $729mm Priced: 2/24/05 Lead: Merrill Lynch Manager: Taberna Capital Management

% of Rating WAL Pricing

Tranche Size Deal (Moody's/S&P/Fitch) (Years) Level

A1 $371,000,000 50.9% --/AAA/AAA 8.3 L + 47

A2 $87,000,000 11.9% --/AAA/AAA 10.1 L + 70

B1 $64,000,000 8.8% --/AA/AA 10.1 L + 110

B2 $10,000,000 1.4% --/AA/AA 10.1 --

C1 $37,750,000 5.2% --/A/A 10.1 L + 180

C2 $25,750,000 3.5% --/A/A 10.1 --

C3 $4,500,000 0.6% --/A/A 10.1 --

D $13,500,000 1.9% --/BBB+/BBB+ 10.1 L + 235

E $37,500,000 5.1% --/BBB/BBB 8.8 L + 315

P/S $77,800,000 10.7% -- -- --

Source: CREDIT SUISSE, Bloomberg, IFRmarkets & MCM

Taberna Preferred Funding II - $1043mm Priced: 6/10/05 Lead: Merrill Lynch Manager: Taberna Capital Management



A1A $400,000,000 38.4% Aaa/AAA/AAA 8.2 L + 43

A1B $106,500,000 10.2% Aaa/AAA/AAA 8.2 L + 43

A1C $10,000,000 1.0% Aaa/AAA/AAA 8.2 --

A2 $86,500,000 8.3% --/AAA/AAA 10.0 L + 65

B $120,500,000 11.6% Aa2/AA/AA 10.0 L + 90

C1 $73,750,000 7.1% --/A/A 10.0 L + 170

C2 $26,000,000 2.5% --/A/A 10.0 --

C3 $15,000,000 1.4% --/A/A 10.0 --

D $31,250,000 3.0% --/A-/A- 10.0 L + 190

E1 $31,750,000 3.0% --/BBB/BBB 9.4 L + 290

E2 $10,000,000 1.0% --/BBB/BBB 9.4 --

F $42,500,000 4.1% --/BB+/BB+ 10.0 L + 500

PS $89,000,000 8.5% -- -- --


Taberna Preferred Funding III - $780mm Priced: 9/14/05

Lead: Merrill Lynch Manager: Taberna Capital Management



A1A $188,500,000 24.2% Aaa/AAA/AAA 8.4 L + 40

A1B (DDraw) $210,000,000 26.9% Aaa/AAA/AAA 8.4 L + 40

A1C $10,000,000 1.3% Aaa/AAA/AAA 8.4 --

A2 $53,500,000 6.9% Aaa/AAA/AAA 10.1 L + 52

B1 $91,250,000 11.7% Aa2/AA/AA 10.1 L + 80

B2 $7,500,000 1.0% Aa2/AA/AA 10.1 --

C1 $36,500,000 4.7% --/A/A 10.1 L + 160

C2 $52,000,000 6.7% --/A/A 10.1 --

D $43,750,000 5.6% --/BBB/BBB 10.1 L + 265

E $31,500,000 4.0% --/BB+/BB+ 10.1 L + 450

PS $55,100,000 7.1% -- -- --


31 March 2006


Appendix B. An overview of trust preferred issuance structure A typical issuance structure for trust preferred securities can be illustrated below:

1) The HC establishes a special purpose subsidiary (Trust), the sole purpose of

which is to issue trust preferred securities. The HC purchases all of the common

stock of the Trust (usually at least 3% of the total capitalization of the Trust).

2) The Trust issues trust preferred securities into the CDOs, or other outside parties.

3) The Trust receives proceeds from the trust preferred securities offering.

4) The Trust uses the proceeds to purchase from the HC long-term junior

subordinated debt with terms matching those of the trust preferred securities.

The Trust distributes interest it receives on the long-term subordinated debt to

pay dividends on the trust preferred securities.

Holdings Company (“HC”)

(Parent Company)

Junior

Subordinated

Debt #4 Proceeds #3

Special Purpose Subsidiary

(Trust) #1

Trust Preferred

Securities #2

Trust Preferred CDO

Proceeds #3


31 March 2006


Bank TruPS: Fine Tuning Historical Bank Failure Rates

211

Rating agencies have traditionally taken a conservative view on bank default rates in

rating bank trust preferred security (TruPS) CDOs, implying a mid-to-low-BBB cumulative

default rate based on historical FDIC intervention rates, with emphasis on the last three

decades of banking history. However, with bank TruPS CDOs performing as well as they

have, we believe the derivation of bank default rates can be fine tuned to imply better

ratings, to more accurately reflect the credit quality of the issuing institutions and the

banking industry in general. In this section, we take a closer look at default rates of bank

trust preferred issuers.212

Since the creation of the agency in 1933, the Federal Deposit Insurance Corporation

(FDIC) has monitored and addressed risks to deposit-insurance funds and limited the

impact of failed bank or thrift institutions on the economy and financial systems. Through

the application of various resolution strategies, the FDIC intervenes on distressed

institutions to insure depositors up to certain statutory limits. Rating agencies and

underwriters approximate bank default rates based on these FDIC interventions.

However, not all interventions result in bank failures and not all failures should be treated

equally. The main issues with existing treatment of bank failures using intervention rates

are two-fold:

1) The inclusion of “Open Bank Assistance” transactions as part of the failure rate;

and,

2) Overemphasis of the banking crisis of the 1980s in deriving failure rates that

reflect the “modern” banking industry (and going forward).

We address each of these issues below, focusing our analysis on commercial banks in

FDIC’s coverage universe.213 We begin with a brief review of the 1980s banking crisis and

the actions that were taken in its wake.

The ‘80s Banking Crisis – Still Applicable?214 By far, the 1980s to early 1990s accounted for the lion’s share of bank failures since the

FDIC’s inception (Exhibit 199). The banking crisis of the 1980s was not caused by a

single event, but rather a combination of forces. These forces included:

• National economic and legislative forces: Volatility in exchange rates of major

currencies in the ‘70s and interest rate variability by the Federal Reserve to

combat inflation challenged the banking industry in the ‘80s. Smaller banks,

which depended on deposit funding, were particularly pressured by rising interest

expenses. Additionally, on the legislative front, the industry saw widespread

deregulation and relaxation of statutory restrictions in an attempt to modernize the

banking industry.

211

This section was originally published in "The CDO Strategist", Issue #10, October 31, 2005. 212

This analysis follows a similar study conducted in 2002 in our Bank Trust Preferred primer. For a general overview of bank TruPS, please refer to "Diversified Bank Trust Preferred CDOs", CSFB CDO Research, October 2003. 213

We exclude savings (thrift) institutions in this analysis. Savings institutions represent a minority of the overall banking industry (about 15% in 2004) relative to commercial banks (85%). Available data on savings institutions is also limited. This study is based on FDIC’s Historical Statistics on Banking, which provides comprehensive lists of individual banks that failed or received financial assistance from the FDIC since 1934. 214

This section makes extensive references to the FDIC publication, "History of the Eighties - Lessons for the Future".

31 March 2006


• Regional recessions: Regional economic stresses resulted in clear geographic

patterns of bank failures. During the 1980-94 period, five states accounted for

nearly 60% of all failures.215 Some of the regional stresses included: severe

downturns related to the collapse of energy prices, real estate related stresses,

especially on the commercial side, and the agriculture recession of the early ‘80s.

• Increased risk with insufficient and untimely oversight: As a result of deregulation

in the ‘80s, banks began taking greater risks without additional supervision to

restrict their discretion and behavior. Any oversight in place was untimely and

infrequent. Examinations of banks declined from 12,300 examinations in 1981 to

8,300 in 1985. Also, the average length of time between subsequent

examinations increased from 13 months in 1979 to 20 months in 1986.

Exhibit 199: Distribution of Commercial Bank Failures, 1934-2004 (by count)

0

25

50

75

100

125

150

175

200

225

1934 1939 1944 1949 1954 1959 1964 1969 1974 1979 1984 1989 1994 1999 2004

Year

Nu

mb

er

of

Fa

ilu

res

$10bn or more

$1bn to $10bn

$100mm to $1bn

Less than $100mm


While some of these forces are outside the control of any regulatory body, the FDIC has

put in place many changes to address the lessons learned from the industry’s most severe

crisis. Some of these changes include:

• Adoption of regulatory capital requirements and risk-based deposit insurance

premiums to make risky behavior less attractive. Prior to 1990, regulators were

limited in their ability to restrain risky lending behavior of profitable banks in the

absence of penalties or costs, which resulted in an abundance of speculative

lending, particularly in commercial real estate.

• Limiting the use of forbearance by requiring more-timely and less-discretionary

intervention of failing banks. While this may result in more closures earlier by the

FDIC, it limits the severity of losses and the potential impact on the rest of the banking

industry.

• Significant improvements in supervision and oversight of banks, and enforcement of

CAMELS ratings. 216 Volume and frequency of examinations have improved

considerably with an annual full-scope examination required since 1991 for most

banks. Additionally, studies have shown that accurate and up-to-date CAMELS

ratings generally identify most of the banks requiring increased supervisory

attention.217

215

By count/occurrence. The five states include: California, Kansas, Louisiana, Oklahoma, and Texas. 216

The acronym “CAMEL” stands for Capital, Assets, Management, Earnings, and Liquidity, five components of a bank’s financial operation that are examined by regulators. In the late 1990s a sixth component was added to the CAMEL rating system, recognizing bank and thrift Sensitivity to interest-rate or market risk (CAMELS). CAMELS ratings are assigned on a scale of 1 to 5 with 1 being the highest and 5 the lowest. “FDIC Banking Review", FDIC. 217

“The Banking Crises of the 1980s and Early 1990s: Summary and Implications”, FDIC

31 March 2006


The banking environment over the last decade suggest that the response to the ‘80s crisis

was fruitful. Since 1994, the banking industry has undergone more than a decade of

relatively benign credit and performance, despite fluctuations in interest rates, turbulence

in the financial markets, and the economic cycle. While it is difficult to predict future

performance and the emergence of new problems with few precedents in the past, we are

encouraged by the changes that have been made since the ‘80s banking crisis and the

solid performance of the industry since.

We believe that the last eleven years represent the “modern” banking era and should be

considered so in the calculation of bank default rates. But more on this, later.

Inferring the Default Rate When a Failure Isn’t a Failure

Since 1934, the FDIC has intervened on commercial banks 2,159 times.218 While most

studies of bank defaults include any intervention as a bank failure, not all interventions are

failures and an ultimate default of obligations. An example of this is the open bank

assistance (OBA) intervention, where a distressed financial institution remains open with

government financial assistance.219

In an OBA, the FDIC seeks to minimize the costs of a failing institution to deposit-

insurance funds. The institution is kept open for public policy motivations, such as

preserving public confidence and maintaining banking services to a community. At the

resolution of an OBA, the bank’s charter continues and creditors are repaid at the expense

of the FDIC, shareholders, and various private sector participants.220 Because ultimate

losses are not realized by debt holders (trust preferred included), this type of intervention

should not be included in approximating the default rate. Since the first OBA transaction

on commercial banks in 1971, there have been 126 instances of OBA interventions. While

the number is small (about 8% of FDIC interventions since 1971), netting out these

transactions does provide a more accurate default picture.

To approximate the default rate using FDIC interventions, we take the following steps:

1) Calculate the number of FDIC interventions, by occurrence and per year, for the

most recent 30-year period from 1975 – 2004, netting out open bank assistance

interventions.

2) Derive the intervention rate for each year, using the total number of outstanding

commercial banks in each year (with failed institutions that year added back on)

as the denominator.

3) Calculate the cumulative intervention rates by summing the rates of each year.

We refer to this as the “failure rate”.

Additionally, we estimate the credit quality by comparing the failure rates to Moody’s

idealized cumulative corporate default rates. The results are shown in Exhibit 200. As

shown, excluding OBA transactions from the failure rate results in a small improvement

(black line compared to dashed-brown line), with the 30-year cumulative rate (11.7%) just

below that of Baa1 corporates (12.4%).

218

Through 12/31/2004, according to FDIC's Historical Statistics on Banking 219

"Managing the Crisis: The FDIC and RTC Experience", Chapter 5 - Open Bank Assistance, FDIC, August 1998 220

There are cases where an institution's charter survives the OBA but is later closed under a different intervention.

31 March 2006


Exhibit 200: Cumulative Bank Failure Rates, 1975-2004

11.690%

0.361%

1.591%

0.0%

2.5%

5.0%

7.5%

10.0%

12.5%

15.0%

17.5%

20.0%

1975

1977

1979

1981

1983

1985

1987

1989

1991

1993

1995

1997

1999

2001

2003

Cu

mu

lati

ve D

efa

ult

Rate

A3 rated corporate Baa1 rated corporateBaa2 rated corporate Baa3 rated corporateCommercial banks (ex. OBA) Commercial banks (inc. OBA)

Source: Credit Suisse, FDIC, Moody’s

Reducing the Weight of the ‘80s Crisis

The curve in Exhibit 200 suggests lower overall failure rates than that of Baa3 corporates

based on curve shape, however we believe the credit quality is better than this. The

primary driver of higher cumulative failure rates in the back-end of the curve stems from

the sharp rise in bank failures during the banking crisis of the 1980s to early-1990s, the

most severely distressed period in US banking history.

An alternative approach, which dilutes the impact of the 1980s’ banking crisis, is to

consider the entire history of bank failures. In Exhibit 201, we show cumulative bank

failure rates from 1934-2004, using every observation available from the FDIC, to generate

the curve. For example, to calculate the 30-year cumulative rate, we take the cumulative

failure rates (as calculated for Exhibit 200) for every consecutive 30-year period since

1934 and derive the average across time. This process is repeated for each period

(1-year, 2-years, 3-years, etc.).

Exhibit 201: Cumulative Bank Failure Rates – All Observations, 1934-2004

5.499%

1.108%2.212%

0.0%

1.3%

2.5%

3.8%

5.0%

6.3%

7.5%

8.8%

10.0%

11.3%

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30Years

Cu

mu

lati

ve D

efa

ult

Rate

A1 rated corporate

A2 rated corporate

A3 rated corporate

Baa1 rated corporate

Commercial banks (all history)


The chart suggests a better credit outlook for banks in terms of failure rates. The curve

(black line) points to lower overall failure rates than Baa1 corporates and flattens towards

the back-end, arriving at a cumulative 30-year failure rate of 5.5%, just under the A1

cumulative corporate default rate of 5.9%.

31 March 2006


However, we note two main problems with this approach, as a result of data shortcomings:

1) Ideally, we’d like to calculate the failure rate based on cohort. Unfortunately, this

data is unavailable.

2) We assume that the banking industry in different periods of time are comparable

to each other, however, we realize that the industry has gone through significant

changes since 1934.

Still, expanding the methodology to include all available bank failures produces a more

level view of the banking industry by diluting the negative effects of the 1980s banking

crisis.

As discussed earlier, many changes and improvements were put in place following the

1980s banking crisis to prevent, or at least to predict, a similar repeat period of distress. If

we take into account only the last 11 years since the end of the banking crisis, which we

consider the “modern” banking era, the resulting failure rates are broadly better.

Using the same methodology as in Exhibit 200, we derive the cumulative bank failure rates

from 1994-2004. As shown in Exhibit 202, the curve suggests lower cumulative failure

rates than A2 rated corporates and similar rates as A1 rated corporates. The higher one-

year rate in 1994 reflects some carry-over affect from the 1980s banking crisis.

Exhibit 202: Cumulative Bank Failure Rates, 1994-2004 – The “Modern” Era

0.6061%

0.2630%

0.00%

0.20%

0.40%

0.60%

0.80%

1.00%

1.20%

1.40%

1.60%

1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004

Cu

mu

lativ

e D

efa

ult

Ra

te Aa3 rated corporateA1 rated corporateA2 rated corporateCommercial banks (ex. OBA)


As we saw in Exhibit 199, small banks (under $100 million in assets) represent the

overwhelming majority of commercial bank failures. According to a recent Fitch study on

bank TruPS CDOs, about 53% of the issuers in 2004-vintage CDOs had assets of $100

million to $1 billion, while issuers with under $100 million in assets accounted for about

8.5%.221 In Exhibit 203, we contrast the failure rates and implied credit quality, across

asset sizes, over two periods in the last two decades: 1) 1994-2004 (which excludes the

1980s banking crisis), and 2) 1984-2004 (which includes the 1980s banking crisis).

221

“Trust Preferred CDO Performance Update”, Fitch, March 1, 2005

31 March 2006


Exhibit 203: Annual Failure Rates for Commercial Banks by Asset Class

1984-2004* 1994-2004

Bank Asset Size

% of issuers in

2004 BTruPS

CDOs

Annual Failure

Rate **

Implied Credit

Quality ***

Annual Failure

Rate

Implied Credit

Quality

Less than $100 mm 8.5% 0.59% Baa2/Baa3 0.069% Higher than A3

$100 mm to $1 bn 53.0% 0.31% A3 0.046% Higher than A3

$1 bn to $10 bn 32.5% 0.26% A2/A3 0.083% Higher than A3

$10 bn or more 6.0% 0.25% A2/A3 0.000% Higher than A3

Weighted Average **** 100% 0.31% A3 0.057% Higher than A3

Source: Credit Suisse, FDIC, Moody’s, Fitch

* 1984 is used as the starting point because data on the number of banks by asset class was only available from this point

** Annual failure rate is calculate as the cumulative rate divided by the number of years

*** Implied credit quality is derived from Moody’s idealized corporate default rates

**** Weighted Average uses Fitch’s 2004 bank TruPS CDO issuer asset compositions as the weights

As shown, commercial bank failure rates imply a weighted average credit rating of A3,

when including the 1980s banking crisis, and even better when excluding, using the actual

breakdown by issuer asset sizes for all 2004 bank TruPS CDOs as the weights. This

suggests that the implied credit quality of bank TruPS CDO collateral may warrant a better

rating than Baa2/Baa3.

Finally, in Exhibit 204, we provide the 5-year, 10-year, and 30-year cumulative commercial

bank failure rates from each of our methodologies above (notice the red boxes in Exhibits

201, 202 and 203 above) since most bank TruPS have either a 5- or 10-year non-call

period and a 30-year legal maturity.

Exhibit 204: Select Cumulative Failure Rates Using Different Methodologies

1975-2004

Most Recent Only (30yrs)

1934-2004

All History (30yrs)

1994-2004

The “Modern” Era (11yrs)

Bank Asset Size Cumulative

Implied

Rating Cumulative

Implied

Rating Cumulative Implied Rating

5-year 0.36% A1/A2 1.11% Baa1 0.26% A1

10-year 1.59% A2/A3 2.21% A3/Baa1 0.61% Aa3/A1

30-year 11.70% A3/Baa1 5.50% Aa3/A1 NA NA


Closing Thoughts The old adage goes: “Only time will tell”. While it remains to be seen whether the banking

industry has learned enough from its past to avoid similar difficulties in the future, we view

the last eleven years of benign credit and low intervention rates as a good start. With a

surge in bank TruPS CDO issuance expected in 2006-2007 (many older bank TruPS are

reaching the end of their 5-year non-call period and are expected to refinance), we think

the asset class is worth considering.

The purpose of this piece is to provide investors with a closer look at one aspect of bank

TruPS CDOs. While our analysis took a less conservative approach than rating agencies,

we also note that there are many other considerations in evaluating the credit quality of

bank TruPS. For example, because TruPS are deeply subordinated obligations, recovery

rates may be very low if defaults occur. That said, the relatively higher average

subordination levels in bank TruPS CDOs should reduce the impact of losses (Exhibit 205).

Also, trust preferred issuance has helped fuel ongoing industry consolidation and M&A

activity. It remains to be seen whether issuers can successfully manage all the risks

associated mergers and growth.

31 March 2006


Exhibit 205: Average Subordination Levels of BTruPS CDOs vs.

Other CDO Types

T rip le- A A verag e Sub o rd inat io n

0%

5%

10%

15%

20%

25%

30%

35%

40%

BTruPS CDO HY CLO CRE CDO Mezz SF CDO HG SF CDO

D o ub le- A A verag e Sub o rd inat io n

0%

5%

10%

15%

20%

25%

30%


Sing le- A A verag e Sub o rd inat io n

0%

2%

4%

6%

8%

10%

12%

14%

16%

18%


T rip le- B A verag e Sub o rd inat io n

0%

2%

4%

6%

8%

10%

12%



31 March 2006


Bank TruPS CDOs: Calling the Underlying222

Common to all trust preferred securities (TruPS) is a non-call feature during the first five or

ten years of the securities’ life, after which the TruPS are callable, typically at par.223,224 In

2006, bank trust preferred securities in early vintage (2001) bank TruPS CDOs will

reach their five-year non-call period and we expect a surge in prepayments and deal

redemptions.

Given that most bank TruPS CDOs are priced to the auction-call date (typically 10 years,

longer than the five-year non-call date of most of the underlying TruPS securities), the

valuation could be misleading if the deal were terminated earlier due to prepayment/call of

the underlying collateral. Due to tightening liability spreads of new-issue TruPS CDOs and

the favorable performance history of bank TruPS, most seasoned bank TruPS CDO

tranches are priced at a premium in the secondary market. It is crucial to assess the

termination date of the bank TruPS CDO accurately, i.e., the expected life of the tranche,

to calculate the fair price of the bond.

In this section, we’ll provide some comments on why it makes sense for banks to call their

early vintage TruPS, review the unique features of bank TruPS CDOs, and provide a

numerical example for bank TruPS CDO investors. We will focus on the 2001 and 2002

vintages of TruPS CDOs, as the collateral in these deals will be exiting their five-year non-

call periods over the next 24 months.

Calling the Collateral – Why it Makes Sense Quite simply, the cost savings for bank issuers are too good to ignore. The cost to fund

trust preferred securities has cheapened significantly since 2001 as a result of solid

performance in the banking sector, significant industry consolidation, investor comfort in

the asset class, and the benign credit environment. 225 Exhibit 206 shows average

collateral cost-of-funding attributes of 2001 and 2002 vintage bank TruPS CDOs versus

that of recent deals (2005). On a weighted-average basis across all pools in the

aforementioned CDO vintages, the cost savings for an issuer that calls its more expensive,

early vintage trust preferred securities and refinances at today’s rates is 161 bps for 2001

TruPS and 123 bps for 2002 TruPS.

Exhibit 206: Average cost savings: too good to ignore

Average Cost Savings

2001 2002 2005* 2005 vs. 2001 2005 vs. 2002

Average WAS (over LIBOR) 384 bps 347 bps 223 bps 161 bps 123 bps

Avg. WAC (for fixed) 9.32% 9.22% 7.76% 156 bps 146 bps

Avg. WA Hybrid (5-yr fixed)** --- --- 6.28%

Avg. % floating rate 90.21% 95.40% 67.85%

Avg. % fixed 9.79% 4.60% 5.70% CD

O C

olla

tera

l

Avg. % hybrid --- --- 26.45%

CD

O

Lia

bili

tie

s

CDO Avg. Cost of Liabilities

(over LIBOR)144 bps 125 bps 65 bps 78 bps 59 bps

Source: Credit Suisse, Moody’s, Fitch

* 2005 features deals backed by both bank and insurance trust preferred securities.

**Hybrid Coupon typically includes a fixed rate for 5-years, followed by a floating-rate coupon.

222

This section was originally published in "The CDO Strategist", Issue #11, November 16, 2005. 223

Nearly all bank TruPS in CDOs feature a five-year non-call, as opposed to ten-year. 224

We note that there are certain TruPS that feature a call premium on the first callable date and the premium decreases on a schedule as the security seasons. However, this is less common. 225

Please see our commentary on bank failure rates, "The CDO Strategist - Bank TruPS: Fine Tuning Historical Bank Failure Rates", Issue #10, October 31, 2005.

31 March 2006


However, the cost savings for bank issuers is likely to be even higher. The figures for

2005 vintage TruPS CDOs in Exhibit 206 include the weighted-average of all pools this

year. Since late 2004, however, most TruPS CDOs have included insurance TruPS

collateral (up to 33% in recent deals), which offer higher spread coupons than banks.

Focusing only on bank issuers, we estimate that the cost to issue bank trust preferred is

more in the range of 150 bps – 180 bps over LIBOR. This suggests a cost savings for

bank issuers that may exceed 230 bps (= 384 bps – 150 bps) if refinanced today,

depending on vintage and issuer credit quality.

Additionally, widespread industry consolidation through mergers and acquisitions (M&A)

may encourage the acquiring issuers to call their outstanding trust preferred securities.

According to Fitch, trust preferred securities in CDOs from the 2001 and 2002 vintages

experienced the highest levels of M&A activity (relative to all vintages from 2000-2004),

with up to 16% of the underlying being acquired as of Q1 2005.226 With the majority of

acquiring banks being large banking institutions ($10 billion or more) with better financial

profiles than the acquirees, it would be economical for the acquirer to retire the more

expensive TruPS (inherited through M&A) and either refinance at cheaper levels or use

alternative forms of funding.

Call attributes unique to bank TruPS CDOs Most trust preferred paper in TruPS CDO pools come from the primary market through

pooled issuance whereby the underwriting process is largely standardized and simplified

across individual banks and each individual bank participating in the pooled issuance

issues trust preferred securities directly into the CDO vehicle, cutting out the costs of

marketing, road showing, etc. Unlike other CDOs, the fact that it is pool-issued gives the

arranger or issuer of the CDO some additional options that are unique to TruPS CDOs.

All of the TruPS in a pooled issue look identical in structure and terms. For example, they

share the same legal maturity, call provisions, default events, etc. These similarities

facilitate the CDO arranger/issuer in executing a pool-wide refinancing of the underlying

securities following the end of the TruPS’ non-call period. The arranger can approach

each bank in the pool and offer to refinance the TruPS into a new CDO vehicle at current

rates. This shares the same effect in terms of terminating the CDO, with the optional

redemption call usually reserved for the equity holder.

Only in the case of collateral that has not yet exited the non-call period or those where

calling may be uneconomical (such as securities requiring significant call premiums) does

the arranger/issuer not have the right to refinance or sell into a new transaction. However,

with the CDO likely de-levered significantly as a result of the majority of the collateral

being prepaid, it might be in the equity holders’ best interest to liquidate the remainder of

the collateral.

Additionally, like the underlying TruPS collateral, bank TruPS CDO liabilities have

tightened in considerably since 2001 as much of the new-issue premium has dissipated

and secondary liquidity has improved for the asset class. The last row of Exhibit 206

shows the spread difference between recent TruPS CDOs and those of 2001 and 2002, a

savings of 78 bps and 59 bps, respectively, for the transaction.

In Exhibit 207, we provide a list of 2001 and 2002 vintage bank TruPS CDOs with a few

attributes.

226

Please see "Trust Preferred CDO Performance Update", Fitch Ratings, March 1, 2005.

31 March 2006


Exhibit 207: 2001 and 2002 vintage bank trust preferred CDOs

Issue Name

Amt

($mm) Pricing Date

% New

Issue TruPS

% 2ndary

TruPS

% Subord.

Debt WAC WAS

%

Fixed % Floating

Static of

Managed

Preferred Term Securities II $347 2/9/2001 97.04% 2.96% 0.00% 8.85% 4.10% 2.96% 97.04% Static

MMCapS Funding I $294 3/21/2001 NA NA NA NA NA NA NA Static

MM Community Funding $525 6/28/2001 100.00% 0.00% 0.00% 10.25% 3.75% 45.00% 55.00% Static

Preferred Term Securities III $516 7/16/2001 93.84% 6.16% 0.00% 8.75% 3.92% 18.09% 81.91% Static

MM Community Funding II $766 11/15/2001 84.62% 5.05% 10.32% 9.79% 3.75% 16.41% 83.59% Static

Preferred Term Securities IV $927 12/4/2001 98.00% 0.00% 2.00% 9.88% 3.60% 1.70% 98.30% Static

Preferred Term Securities V $564 3/14/2002 94.54% 2.83% 2.63% 10.46% 3.60% 2.83% 97.17% Static

MM Community Funding III $540 3/26/2002 94.79% 5.21% 0.00% 9.13% 3.70% 4.63% 95.37% Static

Preferred Term Securities VI $554 6/24/2002 100.00% 0.00% 0.00% n/a 3.45% 0.00% 100.00% Static

TPerf Funding I $492 7/11/2002 94.52% 5.48% 0.00% 9.21% 3.63% 4.47% 95.53% Static

Preferred Term Securities VII $532 9/18/2002 95.01% 0.00% 4.99% n/a 3.40% 0.00% 100.00% Static

TPref Funding II $578 10/16/2002 89.80% 5.27% 4.93% 9.65% 3.45% 0.73% 99.27% Static

Trapeza CDO I $337 10/25/2002 79.53% 20.47% 0.00% 8.34% 3.38% 14.86% 85.14% Managed

TPref Funding III $372 12/11/2002 89.25% 7.93% 2.82% 9.05% 3.34% 6.23% 93.77% Static

Preferred Term Securities VIII $534 12/19/2002 80.19% 9.00% 10.81% 8.68% 3.24% 7.62% 92.38% Static

Source: Credit Suisse, Fitch , Moody’s, trustee reports

Numerical Example Here, we use a sample bank TruPS CDO from 2001 to illustrate the impact of

calling/refinancing the underlying trust preferred securities on the valuation of the CDO.

The CDO is a static deal priced in November 2001 with semiannual payments, the first of

which was in June 2002. Exhibit 208 shows the capital structure and some characteristics

of this sample deal.

Exhibit 208: Sample TruPS CDO

Tranche Rating Coupon Stated Maturity

A Aaa 6M LIBOR + 100 12/15/31

B A3 6M LIBOR + 220 12/15/31

Equity

Auction-Call 10 years End of Non-Call/1st Call December 2006

WAS 6M LIBOR + 375 WAC 9.89%


The composition of the collateral of this deal is very simple, as it was pool issued: all

floating securities share the same coupon spread of 6M LIBOR + 375 bps and almost all

fixed securities share the same fixed coupon of 9.95% with one exception. The non-call

period of the underlying securities and the CDO is the same: December 2006, or five

years after closing.227

Currently, new issue Aaa TruPS CDO is priced at about L+ 33 bps and A3 is priced at

around L+140 bps. If priced to the auction-call date and at these DM levels, Tranche A

would be priced at $103.37 and Tranche B would be priced at $104.14 (see Exhibit 209).

However, as we discussed previously, it is rational to believe, and very likely, that the

entire collateral may be called/prepaid once the non-call period is over. As a result, the

CDO may terminate in December 2006 and thus the tranches should be priced to this date

instead of the auction-call date. Assuming the same DM levels, now the fair prices would

be significantly lower: Tranche A at $100.71 and Tranche B at $100.88.

This illustrates how tranches may be over-valued if priced to the auction-call date if it is

very likely that the CDO could be terminated much earlier – at the end of non-call date.

227

Some deals may have more complicated collateral compositions: more seasoned/secondary trust preferred securities and less homogeneous. The entire collateral pool may not necessarily be called/prepaid at the same time, but the idea and analytics are similar.

31 March 2006


However, even when priced to the end of the non-call (December 2006), we think it is still

a very attractive trade. Take Tranche A as an example: one can earn L+33 bps on a very

short AAA bond with WAL of about 1.1 years.

Exhibit 209: Comparison of valuations: priced to different dates

Scenarios Tranche DM (bps) Fair Price WAL

A 33 $103.37 5.77Priced to 12/2011

B 140 $104.14 6.15

A 33 $100.71 1.10Priced to 12/2006

B 140 $100.88 1.15


Summary In closing, we anticipate a surge in prepayments of early vintage bank trust preferred

securities in the next 24 months, as the securities exit their five-year non-call periods and

the cost savings to bank issuers are very attractive. Furthermore, the unique position of

bank TruPS CDO arrangers/issuers may pave the way for a spike in CDO issuance as

well.

Investors in the asset class should be aware of potential pricing inaccuracies in valuating

bank TruPS CDOs. While future cash flows are effectively cut short by the underlying

collateral redemptions, we believe that vintage TruPS CDO tranches still offer

considerable value given their relatively high spread over a shorter average life.

31 March 2006


Chapter 4: Relative Value and

Secondary CDO Market

31 March 2006


Secondary Valuation Models of Cash Flow CDOs – Review and Pitfalls

228

The secondary CDO market has seen tremendous growth in recent years. However,

valuating seasoned/secondary CDO tranches remains a major challenge confronting CDO

investors. Several valuation models exist in the market with each bearing its own pros and

cons. We review several popular valuation approaches to secondary cash flow CDO

analysis and comment on their pitfalls. We think it is important for investors to understand

the nuances of each model and hope this commentary helps foster the development of

better valuation technologies.

General Approaches to Asset Valuation In finance, regardless of the asset class under analysis, there are generally 3 basic

approaches to valuation:

1. Market comparability approach. This approach derives the value of the target

asset from comparable assets with similar characteristics. An example is real

estate property appraisals, which base target house values on similar houses in

the same neighborhood. This approach is most suitable for illiquid assets.

2. Adjusted discount factor approach. The key to this approach is to determine a

risk premium such that future cash flows are discounted at the risk-adjusted rate.

The resulting present value is the asset value. A simple example is a corporate

bond. We discount future coupon and principal payments at a rate incorporating a

credit spread commensurate with the credit risk for this bond. Obviously, the

challenges with this approach include finding the right risk premium and

separating the liquidity premium from the risk premium.

3. Adjusted cash flow approach. Also known as the “risk-neutral” valuation, the

idea behind this approach is to assign probabilities – of receiving or not receiving

– to future cash flows, in order to make the investor indifferent between investing

in this risky asset and a risk-free asset. If we discount the probability-weighted

cash flows at the risk-free rate, the present value will be the value of the asset.

The key here is to find the “probabilities” – also called “risk neutral probabilities.”

The famous Black-Sholes Model of option pricing is essentially built upon this

approach.

As we will show later, many of the valuation approaches we discuss here can find their

roots in one of these three basic approaches.

Laying Out The Questions First and foremost, we lay out some common questions posed by secondary CDO market

participants in making relative value decisions. These include:

1. Within the same CDO deal, which part of the capital structure offers the best

value?

2. Within the same CDO sector, such as mezzanine SF CDOs, and rating category,

which tranche offers the best value?

3. Across different types of CDOs, which type offers the best value – for example, a

BBB-rated mezzanine SF CDO tranche versus a BBB-rated CLO tranche?

228

This section was originally published in "The CDO Strategist", Issue #14, February 16, 2006.

31 March 2006


In traditional finance theory, such as CAPM, investors make investment decisions by

comparing expected return versus a risk measurement – typically the standard deviation

or variance of the return. Measurements based on this kind of mean-variance analysis,

such as the Sharpe Ratio – calculated as the ratio of excess return over the standard

deviation of return – are widely used, especially in equity investment. For CDOs however,

this approach may not work mainly because of the following two reasons:

1. The mean-variance analysis is based on the assumption that returns follow a

normal distribution. For fixed income securities, especially CDOs, this is hardly

the case.

2. Even if we can assume the normal distribution, there is insufficient return data to

calculate any meaningful risk measurements, given that the CDO market is still

relatively new and information is not very transparent.

In turn, investors have taken alternative approaches to valuation. We discuss these in the

next section.

Existing Valuation Models of Secondary CF CDOs

Approach 1 – Comparing Spreads

Spread comparison is probably the most widely-used and easiest valuation approach.

Two popular methodologies compare spreads across different CDO types, as shown in

Exhibit 210, and relative to historical means, as shown in Exhibit 211 where historical

spreads of the BBB tranche of mezzanine SF CDOs are used.

Based on the credit curves in Exhibit 210, AAA spread levels across different CDO sectors

are similar, while A and BBB spread levels show wider dispersion.229 For example, the

BBB tranche of mezzanine SF CDOs has the widest spread at L+350 bps, followed by

high grade SF CDOs and bank trust preferred CDOs, while BBB CLO tranches offer the

tightest spread at L+180bps. Market participants often make the argument that one asset

type offers relative “spread pick-up” versus another type. In this case, can we argue that

mezzanine SF CDOs offers the best value at BBB level? The answer is not so obvious.

Why? Because this approach fails to address one very important variable – the risk. The

rating only reflects the level of expected default and loss coverage. It does not necessarily

reflect the “volatility” of the default risk and loss rate, i.e., it does not tell how likely the

“realized” defaults/losses will miss the expected numbers. Plus, not all ratings are created

equal – different collateral have different characteristics and different deals have unique

structures.

229

We use new-issue spreads to make our points even though we are addressing issues in secondary valuation. Certainly we can use secondary spreads instead.

31 March 2006


Exhibit 210: CDO Credit Curves by rating (as of the end of January 2006)

0

50

100

150

200

250

300

350

Sr. AAA Jr. AAA AA A BBB

Sp

rea

d (

bp

s)

CLO MZ SF CDO HG SF CDO CRE CDO BTRUPS CDO


Exhibit 211: Historical spreads of BBB tranche of MZ SF CDOs

200

220

240

260

280

300

320

340

360

9/7

/01

12

/7/0

1

3/7

/02

6/7

/02

9/7

/02

12

/7/0

2

3/7

/03

6/7

/03

9/7

/03

12

/7/0

3

3/7

/04

6/7

/04

9/7

/04

12

/7/0

4

3/7

/05

6/7

/05

9/7

/05

12

/7/0

5

BB

B M

Z S

F C

DO

Sp

rea

d (

bp

s)

Historical Average Spread

Hist Avg + 1 Std. Dev.


Comparing current spreads versus historical levels is another popular approach. The

underlying premise for this approach is, over the long run, value – or in this case, spread –

will revert to its long-term mean. Exhibit 211 suggests the BBB tranche of mezzanine SF

CDOs currently looks cheap as its level is higher than the historical average. In addition,

to make the argument more compelling, the level is even higher than the historical

average plus one standard deviation of historical spreads. However, this approach fails to

address some risk factors, such as the risk characteristics of the underlying collateral

changing over time due to changes in collateral composition, and forward-looking factors,

such as future US housing prices.

31 March 2006


Therefore simply comparing spreads should be performed on a first-cut analysis; more in-

depth and rigorous analysis is needed.

Approach 2 – NAV-based Analysis

NAV, or “Net Asset Value”, is probably the most widely used concept in secondary CDO

valuations. The idea is simple: NAV is the market value of the CDO collateral minus any

hedging costs and the outstanding balance of the notes senior to the target tranche.

We show the NAV calculation of a very distressed seasoned SF CDO in Exhibit 212: the

average market value of the underlying portfolio is 90 cents on the dollar. NAV is

expressed in both a dollar amount and as a percentage of the outstanding balance of the

tranche. For the AAA tranche, the $ NAV is the same as the net market value of the

collateral, or 108% of its outstanding balance, while for the BBB tranche the % NAV is only

3%. These “liquidation prices” may be considered as the upper limit of the price for these

tranches. We also show MV (market value) OC which is also often used by traders in the

secondary market.

Exhibit 212: Calculation of NAV

(1) Total Collateral Par $ 278,949,670

(2) Weighted Average Market Price 90.00%

(3) Market Value of Collateral $ 251,054,703 (1) X (2)

(4) Swap $ (10,781,253)

(5) Net MV $ 240,273,450 (3) + (4)

Tranche Name Original Balance Current Balance Rating NAV ($) NAV (%) MV OC

A $ 248,000,000 $ 221,665,195 Aaa $ 240,273,450 108% 108%

B $ 18,000,000 $ 18,000,000 Aa2 $ 18,608,255 103% 100%

C $ 22,000,000 $ 22,000,000 Baa2 $ 608,255 3% 92%

Equity $ 12,000,000 $ 12,000,000


The NAV-based analysis is straightforward and intuitive, however, it has the following

shortcomings: 1) It may be difficult to get accurate collateral market value, especially for

distressed and illiquid assets; 2) The value of the tranche derived from NAV analysis may

not be realistic, because to liquidate the entire deal requires the proceeds be sufficient to

pay down all outstanding notes at par. The rule of thumb is the NAV analysis is relatively

more reliable for first-priority notes.

Approach 3 – Cash Flow-based Analysis

All cash flow-based approaches start with generating future cash flows on the collateral

side based on certain sets of assumptions on parameters such as prepayment, default,

recovery, and reinvestment rate (for managed deals still in the reinvestment period). There

are usually two levels for this exercise:

1. At the asset level. For example, we assign assumptions on each home equity

bond in a SF CDO. If a constant default rate is used, this number is called the

CADR (constant annual default rate) and it is widely used by Wall Street dealers;

2. At the underlying loan level, each home equity loan in the underlying pool for each

home equity bond is assigned certain prepayment, default and recovery

assumptions. The loan-level analysis offers better accuracy, however at the cost

of longer computing time and power.

Once the cash flows of the underlying collateral are generated, the next step is to generate

cash flows of the CDO tranches based on the deal structure and waterfall. Thanks to the

growing usage of analytical packages such as Intex, conducting cash flow analysis is

straightforward and no longer a daunting task.

31 March 2006


Once the cash flows are generated, a price-yield analysis follows: for floating bonds, the

discount margin (DM) – a spread over forward LIBOR curve, at which future cash flows

are discounted to be equated to the price of the bond – is usually used. The higher the DM,

the better the investment at a particular price. Or, if there is a targeted DM, it is used to

calculate the present value of future cash flows; the sum of which is the bond price. This is

essentially the same approach as the “Adjusted Discount Factor Approach” introduced

earlier. However, the question remains: what is the right DM or risk spread to use? The

most common practice is to use the new-issue spread.

Exhibit 213 shows an example of a BBB-rated CLO tranche. The original coupon spread of

this bond is LIBOR + 250bps, while the current new-issue spread is about LIBOR +

180bps. Using 180bps as the DM, the price of this bond is $105.19.230 If this price is

actually traded in the market, then on a mark-to-market basis, this bond has appreciated

about 5 points.

However, this type of analysis has two weaknesses:

1. Simply applying the generic new-issue spread level to all bonds does not capture

the uniqueness of each CDO deal. It seems relatively reasonable for pristine

deals whose underlying collateral have no credit issues but misleading for others.

Even for pristine deals, it would be difficult to concludes that they should all be

trading at the same spread.

2. Between two bonds trading at different DM’s, can we just pick the one with the

higher DM? Unfortunately, there is still something missing: the risk factor.

Exhibit 213: Price-Yield Analysis based on Cash Flows General Information DM (bps) Price ($)*

Type HY CLO 160 106.72

Vintage 2004 180 105.19

Rating Baa 200 103.67

Coupon 3-M LIBOR+250 bps 220 102.17

Current Baa

Spread of HY CLO

(as of Jan 2006)

3-M LIBOR+180 bps 240 100.72

250 100.00

Main Assumptions

Variable Bond Loan

CPR 5% 20%

CDR 2% 2%

Recovery Rate 30% 75%


* Priced to maturity

One popular approach to solve the second issue is to use the so-called sensitivity or

scenario analysis.

In Exhibit 214, we compare two DM profiles against different CADR scenarios: a BBB CLO

tranche versus a seasoned (2004) BBB mezzanine SF CDO tranche. It is harder to simply

pick the bond with the higher DM, as risk factors are considered via different CADRs. The

SF CDO tranche offers higher DM at lower CADR but the DM drops when 4% CADR is

reached – much earlier than the CLO tranche, whose DM – albeit lower – does not drop

until around 9% CADR. The decision to use which CADR is contingent on one’s estimation

on the default risk of the underlying collateral.

230

Also note that the DM at par is the same as the original coupon spread of 250 bps.

31 March 2006


One closely-related concept is the “Break-even CADR”, which is either a default rate

causing the first break in yield/DM – below the coupon yield/spread – or a default rate

resulting in zero yield. The Break-even CADR may be one potential candidate for risk

measurement, similar to using variance to measure risk in traditional portfolio theory. If so,

the reasoning is as follows: given the same expected returns – as measured as DM – the

bond with the higher break-even default rate is more attractive. What if one bond offers a

higher DM but has lower Break-even CADR? Again, the default risk needs to be estimated

for the underlying collateral to answer this question.

Approach 4 – Implied Multiplier Analysis

The Implied Multiplier approach is just one step further than the “break-even” analysis we

just discussed in Approach #3. The first step is to derive the expected default/loss rate,

usually from historical default and recovery statistics of each asset type. The “multiplier”

may be expressed as the ratio of the break-even CADR over the expected default rate.

This approach helps make comparisons across different CDO types.

Aside from all the shortcomings associated with the cash flow-based approach mentioned

previously, the multiplier approach has some additional problems of its own: historical

default rates may not be correctly calculated – yes, even at an aggregate level – and they

may not reflect future risk factors. Another pitfall is, although the break-even CADR is

different for different tranches of a CDO deal, the expected default rate based on historical

experiences is the same for the entire underlying collateral. In other words, within the

same CDO structure, the multiplier of a higher-rated tranche is always higher than that of

lower-rated tranche. Therefore, it can not be used to assess relative value of tranches with

different ratings within the same deal.

Approach 5 – IRR Analysis for Equity Tranches

Equity valuation is arguably one of the biggest challenges in secondary valuation. Internal

Rate of Return (IRR) analysis is the most common way to evaluate an investment in an

equity tranche. The first step is nearly identical to the aforementioned cash flow-based

approach, the only difference is the cash flows of the equity tranche are used. The IRR is

calculated from the resulting cash flows and is usually tested against different assumptions

used on the underlying collateral, such as the CADR.231

As an example, we picked one CLO equity tranche and one mezzanine SF CDO equity

tranche – both issued in late 2004 – and show their equity IRR profiles in Exhibit 215. In

this case, the decision is fairly straightforward: the CLO equity seems to be a better

investment as its maximum IRR is similar to that of SF CDO equity, while also holding up

much better than the IRR of the SF CDO through various CADR scenarios.

231

Same kind of analysis can be found in almost all equity marketing books of new-issue CDOs.

31 March 2006


Exhibit 214: DM versus CADR – comparison of BBB SF CDO vs. BBB CLO tranches

-300

-200

-100

0

100

200

300

400

0 1 2 3 4 5 6 7 8 9 10

CADR (%)

Dis

co

un

t M

arg

in (

DM

, b

ps)

BBB SF CDO tranche

BBB CLO tranche


The result is not surprising, however, as the tight spread environment and shrinking

arbitrage across most CDO asset classes in 2004-2005 forced similar – among CLOs and

HG SF CDOs – baseline IRR’s to around the low teens. If defaults rise in tandem for both

the leveraged loan and ABS (mostly subprime home equity) markets, the higher recovery

rates of leveraged loans suggest CLO equity will out-perform SF CDO equity. In order for

SF CDO equity to be more attractive, it has to have a higher baseline IRR.

The pitfall here resides in the different leverage ratios: it is difficult to compare two IRR

profiles as – when keeping all else equal – the one with higher leverage will always have a

steeper profile. Thus we may have a similar dilemma as the one shown in Exhibit 215. The

same question emerges: how do we balance the trade-off?

Investors sometimes wish to express the value of the equity tranche as the present value

of future cash flows. The challenge again is to determine the right discount rate and the

most common solution is to use the current market expected equity IRR of new-issue

CDOs with similar characteristics as the target deal. One important thing to keep in mind is

that, as time goes by, the value of the equity tranche – expressed as a percentage of its

original balance – will always go down even without any credit deterioration, given its IO-

like nature and pay-down over time. This approach shares all the shortcomings we raised

regarding the cash flow-based approach.

31 March 2006


Exhibit 215: IRR analysis of equity tranches*

-15 .00

-10 .00

-5 .00

0 .00

5 .00

10 .00

15 .00

20 .00

0 0.25 0.5 0 .75 1 1.25 1.5 2 2 .5 3 3.5 4 4 .5

C AD R (% )

IRR

(%

)

E quity of a sample CLO

E quity of a sample M Z S F CDO


* The 2 sample CDOs are both 2004 vintage deals. The recovery rate assumed for CLO is 75% and it is 50% for SF CDO.

Approach 6 – Simulation-based Approach

CADR-based analysis ignores default timing and the impact of correlation. Since the value

of CDO tranches is path-dependent, the timing of defaults and losses could have

significant impact on the final price. Loan-level analysis (for SF CDOs, for example) can, to

some extent, mitigate this issue as the assumed prepayment and default curves would

dictate the timing of bond defaults. However, loan-level analysis considers only one

particular scenario, ignoring the full spectrum of risk factors. While we discussed the use

of scenario or sensitivity analysis to gauge the risk factor, these scenarios could be

arbitrarily specified and may not reflect reality.

As we discussed at the beginning of this section, one of the main valuation approach in

finance is the so-called “Adjusted Cash Flow” approach – or sometimes called “Risk

Neutral Valuation”. The fair value of a CDO tranche can be computed as the risk-neutral

expectation of its discounted cash flows. The key to this type of analysis is to derive the

“market implied” default probability. And the structural complexity and path-dependent

nature of cash CDOs turns our attention to Monte Carlo simulation for a solution.

The main steps for the simulation process can be summarized as follows:

1. Derive market-implied default probabilities/intensities from market prices/spreads

for underlying assets.

2. Specify a dependence structure – i.e., correlation structure – of either asset

returns or default occurrences, and estimate the parameters such as asset

correlations or default correlation.

3. Simulate the default timing of the collateral assets.

4. Run cash flow model to generate discounted cash flows for each default path.

5. Use the discounted cash flows from thousands of paths to estimate the fair value

of the subject bond, including the standard error of the estimation.

This framework works relatively well for CDOs backed by corporate bonds or loans.

However, for SF CDOs or CRE CDOs, it is still an open topic as to how to conduct similar

analysis. Maybe with the development of the CDS market, it will help to derive the market

implied default probabilities of SF and CMBS securities.

31 March 2006


Approach 7 – Re-rating Approach

While the re-rating process is more widely used for monitoring an existing CDO portfolio –

such as managing a CDO^2 deal – it can also be used for relative value analysis.

To re-rate CDO tranches based on an in-house rating system, a typical process may look

like the following:

1. Collect market prices of the underlying assets and calculate the cost/value of

hedges – this is very similar to the NAV approach.

2. Pay close attention to distressed assets – those already rated “CCC” and below,

or assets trading at distressed levels. If necessary, certain haircuts may be

applied, or distressed assets will be assumed to be liquidated at market price.

3. For assets on negative watch list by any rating agency, notch the rating down.

4. Certain reinvestment assumptions have to be made on the principal proceeds and

proceeds from liquidation of distressed assets.

5. If necessary, an in-house criteria can be used to classify the underlying assets

into different industry category, which may not be exactly the same as the rating

agencies’. The difference may cause discrepancy in ratings as different industry

groups have different correlation assumptions.

6. Re-rate the subject bond using rating agencies’ models based on aforementioned

assumptions.

Once the rating is determined, it could be incorporated into the decision process of relative

value and predicting future rating actions.

Approach 8 – Option-like Approach

For many distressed CDO tranches in the secondary market, sometimes it is useful to

treat them like options.

Take a mezzanine tranche of a very distressed SF CDO as an example. The tranche is

PIK-able: as the coverage tests have been breached substantially, all the interest – after

paying the coupon on non-PIK-able tranches – is captured to repay the principal of the

senior most class in the capital structure. We can split this mezzanine tranche into a PO

and an IO piece. In this case, the likelihood of receiving any future interest cash flow is

very low and the IO piece is probably worthless. However, the PO might be worth

something. It can be thought of as an out-of-money call option. Given that the size of these

mezzanine tranches tend to be relatively small compared to the entire deal, it does not

take too much reduction in the realized default/loss rates in order to go from getting paid

nothing to being paid in full in principal. An estimation of the value of this option can be

made if we can figure out the probabilities of various loss rates and then calculate the

probability-weighted principal payment to the bond. The price can be viewed as the

premium for the option. A simulation-based approach can also be adopted.

31 March 2006


Other Considerations in Secondary Valuation There are many other factors investors need to consider when making investment

decisions in the secondary market. We focus on the two important ones: call probability

and the CDO manager’s expertise in managing the CDO.

Throughout our discussion we assumed the CDO tranches are priced to the legal maturity

date. However, there are two very important call features in CDO structures: the optional

redemption by equity investors and the mandatory auction call. The exercise of either call

option can dramatically change the fair value and average life of a bond. Imagine an

investor buys a floating CDO tranche at a premium priced to maturity, if this deal gets

called by the equity holders after the non-call period and the bond gets paid off at par, the

investor will lose the premium paid on the bond. So it is very important to estimate the

probability and timing of the call when evaluating the CDO tranche. We have discussed

the optional redemption and auction call in our previous CDO Strategist and we encourage

our readers to review it.232

The second issue we want to discuss is the CDO manager. Manager performance and

selection are intriguing issues. We believe managers add value to the CDO investment.

However, evaluating a manager’s capabilities and separating the good ones from the

mediocre ones is no easy task, especially for SF CDO managers given the relatively short

history of past performance data. We think the “manager effect” is probably more relevant

for performing and slightly stressed deals, but not so much for very distressed deals, as all

the restrictions and rules embedded in the indenture will probably restrain the manager

from doing anything at all. Nevertheless, we think it is important to take the manager into

the consideration of CDO value.

Closing Thoughts The growing challenge of using the right risk measurement in CDO analysis becomes

increasingly important as the secondary market expands from a handful of players to an

active fixture in the bond markets. Based on our discussion, several potential candidates

for valuation could be considered such as: break-even CADR and the Implied Multiplier of

break-even CADR over the expected default rate. However, each model has its own

pitfalls and nuances. Before improvements and advancements in valuation technologies

can be developed in the future, we encourage investors to fully understand the nuances of

these methodologies and incorporate them into the valuation process.

232

Please refer to The CDO Strategist, May 31, 2005 and The CDO Strategist, July 28, 2005.

31 March 2006


2003 Vintage Mezz. SF CDOs – One of a Kind233

What Makes the 2003 Vintage Special? We think the 2003 vintage of mezzanine SF CDOs offer unique risk-return profiles

because of the following collateral and structural features:234 235

1. Limited exposure to “troubled” ABS sectors;

2. Wider (or Attractive) spreads on both asset and liability sides ;

3. Limited share of non-traditional and higher levered mortgage products such as IO

loans; and

4. Relatively modest concentration of residential mortgage loans originated in higher

price growth areas.

On a risk-adjusted basis, we think 2003 vintage of mezzanine SF CDOs are attractive to

secondary CDO investors. Exhibit 216 details a general profile of the 2003 SF CDO deals

used in our analysis. We discuss each point in more detail.

The underlying collateral – limited exposure to troubled ABS sectors In CDO investing, picking the right asset classes is the most important step. It is well-

known that many older vintage mezzanine SF CDOs (1999-2002) suffered downgrades

and losses because of significant exposure to “troubled” sectors such as manufactured

housing, aircraft leasing, etc.

Since 2003, the diversity of SF CDO collateral has decreased and there has been a shift in

the composition - from highly diversified pools with “troubled” sectors to more concentrated

pools with mortgage-related assets. This trend is illustrated in Exhibit 217; most 2003

vintage deals don’t have significant exposure to sectors such as MH and aircraft leasing,

with the exception of deals such as Deal 13.236 Before 2003, on average, SF CDO

collateral included around 10% MH and 4-5% aircraft leasing.

On the other hand, for most of the 2003 vintage mezzanine SF CDOs, the exposure to

residential mortgage-related assets, residential B&C mortgage and home equity and

residential A mortgage combined, jumped significantly.237

As shown in Exhibit 217 more than half of the deals have greater than 50% exposure to

residential B&C and home equity collateral. 238 Combined with residential A, the total

exposure could reach close to 90%, such as Deal 6. In earlier vintages, the share is

around 20% on average.

233

This section was originally published in "The CDO Strategist", Issue #5, July 15, 2005. 234

The SF CDOs discussed include only mezzanine, multi-sector SF CDOs, excluding high-grade SF CDOs, CRE (or CMBS) CDOs, and CDO-Squared. 235

The term, "2003 vintage mezz. SF CDOs", represents a group of CDO deals with specific characteristics and not just a time frame. For example, if a deal is done in late 2002 or early 2004 but contains similar characteristics as discussed, our arguments can be applied to it as well. 236

To date, Deal 13 is the only one that has been downgraded by any of the rating agencies. 237

We combine residential B&C mortgage and home equity in our discussion as there was some confusion in the market regarding the classification of these two sectors, and so we use “home equity” for convenience. Please see CSFB special report, “Classification Conundrum: Residential Mortgage Classifications in SF CDOs”, December 23, 2004. 238

This number could be even higher for more recent deals, such as 2004 and 2005 vintages.

31 March 2006


Exhibit 216: General Profile of 2003 Vintage SF CDOs*

Deal Closing Date

Static/

Managed

Minimum

Diversity-Score Original WAR Turbo

Equity

Cap

Reinvestment

Period

Non-Call

Period

Deal 1 May-03 Managed 18 Baa1/Baa2 Y 18.0% 48 48

Deal 2 Nov-03 Managed 19 Baa1/Baa2 48 48

Deal 3 Jul-03 Managed 16 Baa1/Baa2 Y 23.0% 36 36

Deal 4 Sep-03 Managed 25 Baa2/Baa3 N 42 60

Deal 5 Jan-03 Managed 16 A3 Y 36 36

Deal 6 Aug-03 Managed 15 A3/Baa1 36 36

Deal 7 Aug-03 Static 16 Baa1 Y 18.0% 0 36

Deal 8 Oct-03 Managed 20 Baa2 N 48 48

Deal 9 Jul-03 Static A3 Y 0 108

Deal 10 Jun-03 Managed 22 Baa3 Y 10.0% 36 36

Deal 11 Jan-03 Managed 21 Baa2/Baa3 Y 25.0% 48 36

Deal 12 Jun-03 Managed 20 Baa1/Baa2 N 48 48

Deal 13 Dec-03 Baa1/Baa2 Y 0

Deal 14 Nov-03 Managed 24 Baa2/Baa3 Y 14.0% 24 36

Deal 15 Nov-03 Managed 16 Baa1/Baa2 Y 23.0% 36 36

Deal 16 Dec-03 Managed 15 Baa1/Baa2 Y 16.0% 36 36

Deal 17 May-03 Managed 18 Baa1/Baa2 36 36

Deal 18 Feb-03 Managed 18 Baa1/Baa2 Y 20.0% 36 36

Deal 19 Jul-03 Managed 20 Baa2 48 48

Deal 20 Oct-03 Managed 20 Baa1 60


* The SF CDOs listed here only include mezzanine, multi-sector SF CDOs, while excluding high-grade SF CDOs, CRE (or CMBS) or real estate CDOs, and CDO-Squared.

The macro economic condition – high spread environment In 2003, HEL spreads were at their widest levels in the last 5 years. As shown in Exhibit

218, 5-year BBB spreads started widening in the second half of 2002, reaching a high of

LIBOR + 350 bps, before retreating to 160-170 bps in the second half of 2003. We believe

most of the tightening is attributed to the CDO bid; most deals in Exhibit 216 closed in the

second half of 2003.239

As most HEL bonds in 2003 vintage SF CDOs are issued in late 2002 or in 2003, the wide

spread provides an attractive return on the asset side of a CDO. In addition, the liability

spread of SF CDOs was similarly wide, in tandem with the asset side. However, the

arbitrage spread was also at its historic high for SF CDOs (Exhibit 218). Based on

CREDIT SUISSE’s excess spread measure, the Multi-Sector Arbitrage Pointer (or MAP),

as shown in Exhibit 219, the arbitrage spread for equity holders was twice the current

level.240

Wider spreads on the underlying assets also reduced the incentive for CDOs to move

down the credit spectrum to Baa3 and below, which represent greater risks, such as rising

interest rates and slowing-down in the housing market.

239

It normally takes 6 to 12 weeks between warehousing and deal closing. 240

Please note MAP does not consider default and is only for indicative purposes.

31 March 2006


Exhibit 217: Original Collateral Allocation of 2003 Vintage SF CDOs

Deal

Residential

B&C

+ Home Equity Residential A MH

Aircraft

Leasing CBO CMBS Corporate

Auto &

Credit Card

Deal 1 54.8% 10.67% 1.85% 1.3% 5.1% 3.8% 3.4% 16.9%

Deal 2 57.8% 14.3% 0.0% 0.0% 8.6% 9.1% 3.8% 4.9%

Deal 3 46.4% 15.5% 5.1% 0.0% 4.9% 9.1% 0.0% 10.4%

Deal 4 31.5% 14.6% 3.3% 1.7% 7.5% 13.6% 2.0% 28.0%

Deal 5 56.9% 28.5% 9.8% 0.0% 0.0% 0.3% 3.0% 0.0%

Deal 6 59.5% 27.4% 1.8% 0.0% 1.8% 1.7% 0.0% 0.0%

Deal 7 55.0% 15.9% 0.0% 0.0% 1.0% 7.4% 0.0% 9.2%

Deal 8 58.8% 11.0% 0.0% 0.0% 5.6% 16.9% 0.8% 3.7%

Deal 9 38.9% 19.7% 5.0% 1.0% 5.1% 15.0% 1.0% 9.8%

Deal 10 54.6% 14.6% 7.8% 0.0% 5.0% 10.3% 1.2% 0.0%

Deal 11 49.8% 10.2% 6.7% 6.6% 8.3% 5.7% 6.0% 9.1%

Deal 12 37.7% 16.8% 2.3% 0.8% 10.9% 10.1% 0.0% 14.7%

Deal 13 40.9% 6.8% 12.1% 5.1% 10.3% 8.8% 0.0% 14.1%

Deal 14 54.7% 11.0% 0.8% 0.9% 1.5% 13.1% 7.3% 7.1%

Deal 15 54.3% 5.4% 0.0% 0.0% 22.8% 1.1% 0.0% 2.53%

Deal 16 56.2% 18.2% 0.0% 0.0% 4.9% 10.0% 0.0% 4.3%

Deal 17 57.1% 9.4% 4.1% 0.0% 3.6% 4.0% 7.9% 3.1%

Deal 18 47.2% 13.1% 6.9% 0.0% 4.0% 13.7% 0.0% 1.9%

Deal 19 50.4% 15.6% 0.0% 0.0% 7.0% 7.9% 0.0% 3.4%

Deal 20 32.7% 7.8% 0.5% 1.1% 9.6% 33.6% 8.8% 4.1%


Limited share of non-traditional and more levered mortgage products Compared to newer vintages HELs, HEL deals issued in 2003 do not have as large a

share of IO (interest only) loans. Exhibit 217 shows the shares of IO loans of each vintage

at an aggregate level. It is clear that in 2004 and 2005, the share of IO loans jumped

dramatically: from 1-4% in 2002 and 2003 to 14% in 2003 and 21% in 2005, based on the

composite.

For investors concerned about high concentration risk of IO loans, 2003 vintage contains

less exposure. Given the short history and limited empirical evidence of this product type,

a track record of performance is still being established.241

In the prime mortgage and Alt-A universes, we see similar patterns. Although the IO share

in these asset classes had already jumped to high levels in 2003, it is not as high as 2004

or later.242

241

For a detailed discussion on IO loans, please refer to CSFB special report, "Subprime Interest-Only Loans: Attributes and Early-Stage Performance", January 2005. 242

Please refer to CSFB special report, "Spotlight on Interest-Only Loans: Prime and Alt-A Fixed-Rate MBS", May 2005.

31 March 2006


Exhibit 218: BBB HEL Spread vs. SF CDO Liability Spread

0.00

10.00

20.00

30.00

40.00

50.00

60.00

70.00

80.00

90.00

Jan-00 Jul-00 Jan-01 Jul-01 Jan-02 Jul-02 Jan-03 Jul-03 Jan-04 Jul-04 Jan-05

Date

SF

CD

O L

iab

ilit

y S

rea

d (

bp

s)

0

50

100

150

200

250

300

350

400

5-Y

ea

r H

EL

Sp

read

ov

er

LIB

OR

(b

ps

)

SF CDO Aggregate Liability Spread 5-Year HEL Floating Spread


Exhibit 219: Multi-Sector Arbitrage Pointer (MAP)

0

20

40

60

80

100

120

140

160

8/31/01 2/28/02 8/31/02 2/28/03 8/31/03 2/29/04 8/31/04 2/28/05

MAP


Relatively modest concentration of subprime loans generated in high HPA areas than more recent HEL deals Given the dramatic increase in US housing prices, the concentration of loans generated in

higher Home Price Appreciation (HPA) areas rose significantly recently. Such a

concentration may represent higher risks in the face of a housing market slow-down. Going

forward, we believe a slower growth rate of housing prices could increase the loss severity

rate of HEL loans significantly, as well as increase default and delinquency rates.

31 March 2006


Exhibit 220: Share of IO Loans*

Vintage

2001 2002 2003 2004 2005

Composite 0% 1% 4% 14% 21%


* These figures are directly from CREDIT SUISSE’s “Subprime HEAT Update”, June 2005.

Exhibit 221: HPI % Change Annualized Quarterly (=Quarterly Growth*4)

0%

5%

10%

15%

20%

25%

1997Q1 1997Q4 1998Q3 1999Q2 2000Q1 2000Q4 2001Q3 2002Q2 2003Q1 2003Q4 2004Q3

Source: Credit Suisse, OFHEO, Bureau of Labor Statistics

As shown in Exhibit 221, the Housing Price Index (HPI) jumped dramatically around the

end of 2003 to close to 14.4% and further up to 19.2% in 2004. Based on this evidence,

we believe most HEL deals issued in late 2002 and 2003 should have moderate

concentration of loans generated in high HPA areas relative to more recent deals, and

thus less risk for 2003 vintage mezzanine SF CDOs.

Implications for secondary valuation Most secondary valuations should be done on a deal-by-deal basis, especially for SF

CDOs given their heterogonous nature. However, the 2003 vintage possesses some

unique characteristics. These attributes we discussed above have significant implications

for secondary valuation and sensitivity analysis.

Characteristics such as wider spreads on CDO liabilities imply higher return for CDO

investors, holding everything else equal. Some factors, such as low exposure to troubled

ABS sectors and low share of IO loans, imply lower risks in potential adverse scenario.

Therefore, on a risk-adjusted basis, we think the 2003 vintage is attractive.243 For example,

based on Moody’s index, the mezzanine OC cushion of 2003 vintage is the most stable

among all vintages. Similar to wine collecting, this is one vintage you may not want to miss.

243

As we emphasize repeatedly, these conclusions are on an aggregate level. For certain deals, such as Deal 13 which is more like an older vintage deal, the performance profile may not be similar to the rest of the vintage.

31 March 2006


Finding Value in Senior Tranches of Distressed SF CDOs

244

Since early 2004, SF CDO downgrades have increased. Most downgrades are from early

vintage SF CDOs – the 1999 to 2001 vintages and select 2002 deals. The main driver for

these downgrades is the poor performance of certain ABS sectors such as manufacturing

housing (MH), aircraft leasing and franchise loans.

We find 78% of 2000 vintage SF CDOs experience at least one downgrade; 58% and 30%

of 2001 and 2002 vintages, respectively, have been downgraded.245 This doesn’t suggest

bad news for all the SF CDO investors because it depends on the tranche. We believe the

senior tranches from early vintages of SF CDOs may offer attractive value, especially if

they are trading at a discount. We analyze an actual transaction to illustrate this

opportunity. The deal was issued in 2001 and is currently failing all performance tests.

Exhibit 222: Distribution by Asset Type (Based on Balance as of 4/29/2005)

Credit Card

6%

Home Equity

18%

NA

7%

Receivables

6%

Recreation Vehicle

2% RMBS

11%

Manufactured

Housing

15%

Franchise

4%

Equipment

1%

CMBS

14%

CDO

2%

Auto Loans

3%Airplane

11%


Exhibit 222 shows the asset allocation of this deal as of 4/29/2005. About 30% of the

collateral is in a combination of MH, aircraft leasing and franchise loans. Exhibit 223

shows the rating distribution; the share of below-Caa1 is about 13%. Undoubtedly, this is

a very distressed deal.

244

This section was originally published in "The CDO Strategist", Issue #3, June 15, 3005. 245

Based on deal count and as of 6/10/2005.

31 March 2006


Exhibit 223: Distribution by Ratings (Based on Balance as of 4/29/2005)

A1

1%

A2

5%Aa3

1%Aaa

3% B1

4%

B3

8%

Ba2

5%

Baa1

8%

Baa2

26%

Baa3

18%

NA

3%

C

4%

Ca

3%

Caa3

4%

Aa2

2%A3

0%

Caa1

0%

Caa2

2%


To generate the cash flows, we apply asset-level prepayment, default and recovery

assumptions. One of the biggest challenges of evaluating seasoned SF CDO deals is to

come up with these assumptions for esoteric and off-the-run asset types such as aircraft

leasing and franchise loans. We rely on our internal models and expertise to generate

prepayment, default and recovery assumptions for home equity, MH, RMBS, CMBS, auto,

and credit card deals. For franchise loans or aircraft leasing, we used conservative

assumptions: for franchise loans, we use a prepayment speed of 30% CPR, a default rate

of 20% CDR, and a severity rate of 60%; for aircraft leasing, we assume no prepayment, a

default rate of 80% CDR246 and a severity rate of 80%.247

By running these assumptions through Intex, we generate the cash flows for each tranche.

Exhibit 224: Capital Structure of the Sample Deal (as of 4/29/2005)

Tranche

Original

Balance

Current

Balance

Original

Moody's

Current

Moody's Coupon OC Target IC Target

A 177,500,000 100,843,494 Aaa Aaa L+48

B 50,000,000 50,000,000 Aa3 Aa3 L+100 105% 117.50%

C 12,500,000 11,952,287 Baa2 Baa2 L+168 102% 107.50%

Equity 12,000,000 12,000,000


Exhibit 224 shows the capital structure, the ratings of each tranche, and the OC/IC targets.

This deal has an embedded Turbo structure: the return on the equity tranche is capped at

26.75% and any remaining interest proceeds are then used to pay down Classes C, B and

A, in that order. This explains why Tranche C has paid down some principal while Tranche

B has not. Also, the deal has exited the reinvestment period.

Because of failing OC/IC tests, both interest and principal proceeds are redirected to pay

down Tranche A until the tests are satisfied. As Exhibit 225 and Exhibit 226 show, the IC

and OC tests will never be cured, based on the projected cash flows.

246

Most aircraft leasing bonds are rated CCC in this deal. 247

Almost all of these franchise loans and aircraft leasing deals are not modeled by Intex. As a result, the assumption provided will be applied at the bond level, instead of at the underlying collateral level.

31 March 2006


Because of the de-levering effect,248 even for a severely distressed deal like this, the

Tranche A can still provide value. As Exhibit 227 indicates, even under very conservative

assumptions, if traded at par, Tranche A can offer a discount margin of 48 bps over 3-

Month LIBOR, which is the same as its coupon spread. If traded at discount, this bond

represent more attractive returns. For a bond with a WAL of 1.54 years, we view this is a

good investment and believe many other senior tranches of seasoned and distressed SF

CDOs may offer similar opportunities.

Exhibit 227: Price/Yield Table for Tranche A of Sample SF CDO

Price Yield (%) Discount Margin (bps)

99 5.2271 119

99.25 5.0524 101

99.5 4.8785 83

99.75 4.7054 66

100 4.5332 48

100.25 4.3619 30

100.5 4.1914 13

100.75 4.0217 -5


248

The subordination level at the beginning of our analysis is about 42%, based on Table 1. As Class A continues to de-lever, the subordination level will get higher.

Exhibit 225: Projected Class B IC Test Exhibit 226: Projected Class B OC Test

0%

20%

40%

60%

80%

100%

120%

140%

Jan-04 May-05 Oct-06 Feb-08 Jul-09 Nov-10 Apr-12 Aug-13 Dec-14 May-16

IC R

atio

Threshold Actual

0%

20%

40%

60%

80%

100%

120%

Jan-04 May-05 Oct-06 Feb-08 Jul-09 Nov-10 Apr-12 Aug-13 Dec-14 May-16

OC

Rati

o

Threshold Actual

Source: Credit Suisse, Intex Source: Credit Suisse, Intex

31 March 2006


Seasoned Senior CLOs Should Trade Even Tighter

249

The Idea In the secondary market, many seasoned AAA-rated CLO bonds are trading at the same

level as new-issue bonds, which stands at around LIBOR plus 25 bps. We think many

bonds nearing the end of the non-call period (less than two years) with clean collateral

should be trading at tighter levels.

Negative Basis Trade A negative basis trade occurs when the bond spread is trading wider than the credit

default swap (CDS) spread (the cost of protection): one can capture the net spread by

going long the cash bond and simultaneously hedging out the credit risk by buying

protection through a CDS contract. Currently, the spread of a CDS on a AAA-rated CLO

bond is a little under 10 bps. Investors can hedge out the counterparty (of the CDS) risk by

buying additional protection against the counterparty’s default risk. If an investor buys the

AAA bond offered at 25 bps and buys protection from a CDS at around 10 bps and

additional insurance against counterparty risk, the investor is locking in a near-risk-free

return of 13-15 bps over the next 7.5 to 8.5 years, the average life of typical new-issue

AAA CLO bonds.

What Does This Mean For Seasoned AAA CLO Bonds? We think seasoned AAA-rated CLO bonds, with less than or equal to two years remaining

in the non-call period, should trade close to the low-teen level, or at least tighter than the

25 bps level priced to the first call date. Our reasons for this belief are as follows:

1. Given that the first call date is approaching within two years, these are very short

bonds if called. Compared to a 7.5- to 8.5-year bond at 13-15 bps, a two-year or

shorter bond at 25 bps is evidently very attractive.

2. From a credit perspective, most of these CLO deals are performing well. Under

normal conditions, it is unlikely that an AAA-rated bond from these currently well-

performing deals will result in losses in a very short time period.

3. Even if these bonds are not called, it could be even better, assuming the credit

situation does not deteriorate disastrously, as we will explain later.

An Example To back up our point, we use a real CLO deal as an example. Exhibit 228 shows the

detailed information of this deal – a 2002 vintage CLO whose non-call period will end in

two years. This deal, along with almost all the CLO deals in the same vintage, is

performing well: it is passing all performance tests and the market value of the collateral is

101.4%.250

249

This section was originally published in "The CDO Strategist", Issue #4, June 29, 2005. 250

Please refer to Moody's Deal Score Report, June 2005.

31 March 2006


Exhibit 228: Sample CLO Deal

Deal Information

Issue Date 10/1/02 WAC of Fixed Assets 9.08%

Reinvestment End Date 7/15/07 WAS of Floating Assets 3.03%

Non-Call End Date 7/15/07 Floating Rate Assets 77.42%

Legal Maturity 10/15/16 Payment Frequency Quarterly

Total Size 450,000,000 Market Price of Collateral 101.4%

Capital Structure

Tranche Name Current Balance Spread/Coupon Rating

Expiration Date of Make-Whole

Premium

A 346,000,000 L+44 bps Aaa N/A

B1 15,250,000 L+140 bps A2 N/A

B2 24,000,000 7.045% A2 4/15/2012

C1 11,000,000 L+240 bps Baa2 N/A

C2 6,000,000 8.055% Baa2 7/15/2012

D 13,000,000 12.78% Ba2 1/29/2012

Equity 34,750,000 N/A


Prepayment of HY Loans 15% CPR Prepayment of HY Bonds 5% CPR

Default of HY Loans 0.5% CDR Default of HY Bonds 2% CDR

Recovery of HY Loans 70% Recovery of HY Bonds 30%


Using our call option model, we can determine whether a deal will be called or not on each

call date from an economic perspective, and calculate the price at each call date as well

as the price if never called.251

Exhibit 229: Cash Flows on the First Redemption Date

Value

Asset Notional (on redemption date) $450,578,483

Market Price of Assets* $101.40

Market Value of Assets (on redemption date) $456,886,582

Swap Termination Payment (on redemption date) ($1,218,128)

Principal and Premium to Liabilities ($421,288,570)

Cash flow to Equity (on redemption date if called) $34,379,884

IRR of Equity** 12.00%

PV of cash flow to equity on call date $27,052,134

PV of cash flow to equity before call date $16,603,442

Total PV of cash flow to equity if called $43,655,576

Cash flow to Equity (on redemption date if not called) $37,417,291

Total PV of cash flow to equity if not called $46,045,588

To Call or Not to Call? Not Call


* Assume dirty price with accrued interest for simplicity

** The current IRR available to equity holders from alternative investments

251

For a detailed discussion on CDO call options, please see our The CDO Strategist (Issue #2), May 31, 2005.

31 March 2006


Exhibit 230: Results of Base Case on the First Redemption Date

Tranche Name

A B1 B2 C1 C2 D

(1) WAL of Tranche if no call 4.37 6.8 6.79 7.05 7.04 7.65

(2) WAL of Tranche if called 2.04 2.04 2.04 2.04 2.04 2.04

(3) Remaining Maturity to Make-Whole Expiration Date (Year) 5.0 5.25 4.75

(4) Make-Whole Spread (bps) 96 148 389

(5) Treasury Rate for Make-Whole Premium (based on (3)) 3.69% 3.69% 3.66%

(6) Treasury Rate for Pricing if Called (based on (2)) 3.58% 3.58% 3.58%

(7) Treasury Rate for Pricing if No Call (based on (1)) 3.69% 3.70% 3.70%

(8)

Pricing Spread over LIBOR (for floating)/over

Treasury (for fixed) 0.25% 0.75% 1.20% 1.80% 2.20% 5.15%

(9) Coupon Rate (Fix)/LIBOR Spread (Float) 0.44% 1.40% 7.05% 2.40% 8.06% 12.78%

(10) Make-Whole Premium 10.64% 13.21% 20.71%

(11) Remaining Notional (on Redemption Date) 346,000,000 15,250,000 24,000,000 11,000,000 6,000,000 13,000,000

(12) Optional Redemption Payout = (11)*(1+(10)) 346,000,000 15,250,000 26,554,593 11,000,000 6,792,326 15,691,651

(13) PV of Optional Redemption Payout 318,594,460 13,901,979 24,145,315 9,819,083 6,055,348 13,201,475

(14) PV of cash flow before call 28,718,781 1,541,284 3,206,626 1,308,555 906,618 3,018,718

(15) Total PV of cash flow if called 347,313,241 15,443,263 27,351,941 11,127,638 6,961,966 16,220,192

(16) Price if called $100.38 $101.27 $113.97 $101.16 $116.03 $124.77

(17) Price if not called $100.79 $103.91 $112.69 $103.57 $112.63 $122.07


Priced at $100.38 to the first call date, the A tranche offers a discount margin (DM) over

forward LIBOR of 25 bps with an average life of two years (see Exhibit 230). As discussed,

this is attractive compared to where the negative basis trades are being done. We think it

should be trading tighter. If priced to the first call date at 13 bps DM, the price should be

$100.61.

The above discussion is based on the assumption that the deal will be called on the first call

date. How likely is it that this deal will actually be called? Based on the baseline assumptions

and our model, it turns out that this deal will not be called on the first call date (see Exhibit

229).252 The good news is that the later the deal is called, the better. As a matter of fact, for

floating bonds, as the call date moves closer to the maturity date, the price if called will

converge up to the price if never called. As it turns out, economically, this deal should be

called on 4/15/2009. If it does get called on this date, the AAA bond at $100.38 will give the

investor a DM around 32 bps for a bond with 2.9-year average life.

Conclusion The risk is that the deal does not get called and something disastrous happens causing a

loss on the AAA bond. However, even in our most stressed scenario,253 the AAA bond did

not suffer any loss and still looks cheap at the 25 bps level. We believe seasoned AAA

CLO bonds close to the end of the non-call period provide good value at the current

pricing of LIBOR plus 25 bps.

252

Faster prepayment speeds and higher market price will increase the probability of call. 253

For HY bonds: CDR at 15 and recovery at 25%; for HY loans: CDR at 8 and recovery at 50%. The CDRs used here are historical highs since 1992 and the recovery rates are historical lows since 1995. For more details on historical rates, please see CSFB’s Leveraged Finance Research, “An Introduction to Cash Flow CLOs”, May 3, 2005.

31 March 2006


Junior AAA of HG SF CDOs Offers Attractive Value

254

Most investors are required to adhere to certain investment guidelines and eligibility

criteria, which can vary by credit rating, sector, investment horizon and more. Therefore,

we discuss relative value by risk/return profiles. And in this issue’s Strategy section, we

focus on investors seeking AAA-rated assets with 7-9 year weighted average lives. We

think the junior AAA tranches of high grade SF CDOs present some very attractive

opportunities.

In a typical HG SF CDO, there are usually two (or more) AAA-rated tranches: the senior-

AAA and the junior-AAA. The senior-AAA usually accounts for 70%-90% of the deal and

is funded either as short-term notes, such as ABCP or money-market tranche, or as term

notes. The junior-AAA usually accounts for 5%-6% of the CDO and is offered at a higher

spread, with an average life from 7 to 9 years.

We think the junior-AAA’s of HG SF CDOs are attractive for the following reasons:

1. They offer attractive spread pick-up over almost all other AAA bonds in the

primary markets of structured finance. Currently, the spread of junior-AAA’s of

HG CDOs is around 45 bps. The all-in funding cost of ABCP tranches is around

L+24 bps while the spread on term-funded senior-AAA’s is around 27 bps in HG

SF CDOs. As shown in Exhibit 231, the spread pick up of junior-AAA over senior-

AAA is around 20 bps. Compared to the AAA bonds in other sectors, as shown in

Exhibit 232, junior-AAA’s in HG SF CDOs offer more spread as well.

Exhibit 231: Junior AAA Spread vs. Senior AAA Spread*

10

20

30

40

50

60

70

80

90

Fe

b-0

4

Ap

r-0

4

Ju

l-0

4

Se

p-

Oct-

04

No

v-0

4

No

v-0

4

No

v-0

4

Fe

b-0

5

Ma

r-0

5

Ap

r-0

5

Ju

l-0

5

Ju

l-0

5

Au

g-0

5

Sp

rea

d (

bp

s)

10

15

20

25

30

35

40

45

50

55

60

Diffe

ren

ce

(bp

s)

Snr AAA Spread Jnr AAA Spread Difference


* For ABCP tranches, we use a float all-in spread of 24 bps.

254

This section was originally published in "The CDO Strategist", Issue #7, September 15, 2005.

31 March 2006


Exhibit 232: AAA Spreads of Select Structured Products

Sector AAA Floating Spread (over LIBOR, bps)

HY CLO AAA 25

Bank Trust Preferred CDO Junior AAA 43

Mezz SF CDO Junior AAA 45

HG SF CDO Junior AAA 45

CRE CDO Junior AAA 39

5-year HEL (Float) AAA 23

7-year Credit Card (Float) AAA 7

10-year CMBS (Fixed) 27*


* Over swap

2. Sufficient loss coverage

Because of the higher credit quality of the underlying pools in HG deals, loss

coverage ratios of most junior-AAA’s are sufficient enough to withstand principal

losses for the given rating.

Similar to Exhibit 38 (see Insight section), we can use the expected loss rates

derived from Moody’s impairment rates and calculate the loss coverage ratios of

junior-AAA tranches. The results are in Exhibit 233. Most of the coverage ratios

fall in the range of 13 to 20 times, sufficient to cover potential losses.

Another check is to use “Break-even Default Rates”, defined here as the annual

default rate resulting in a break in yield (for floating bonds, a break in discount

margin, i.e. a discount margin below the coupon spread of the notes). We

compare the Break-even Default Rates to the annual default rates of each deal,

calculated from the expected loss rates derived in Exhibit 38, assuming a

recovery rate of 55%.255

As shown in Exhibit 234, the ratio of Break-even Default Rate over the implied

annual default rate is very high for most of the deals: at least 20 times in a faster

prepayment scenario.256 To put these numbers into perspective: for a resulting

loss on the junior-AAA tranche, the annual default rate has to be at least 20 times

higher than the empirically implied default rate! We view this as a very remote

event. That said, the existence of a high coverage ratio does not entirely

eliminate the possibility of a loss on the tranche. This analysis is based on pool-

level assumptions; more robust asset-level analysis is needed for further

investigation.257

In summary, we think the junior-AAA tranche of HG SF CDOs offer an attractive risk/return

profile. Investors of AAA-level risk should take a close look at this asset class and explore

potential investment opportunities. As shown in Exhibit 231, junior-AAA spreads have

followed a tightening trend and we believe given a robust risk/return profile, spreads

should remain tight or continue to grind in. Furthermore, the spread between senior-AAA

and junior-AAA is likely to converge. We note that there could be downgrade risk on the

junior AAA tranches in extreme adverse situations.

255

55% is the recovery rate (45% severity rate) we used to calculate the expected loss rates. We also use the forward LIBOR curve. 256

When prepayment is faster, it is less likely to suffer greater losses from back-loaded defaults, and thus more likely to have a higher break-even default rate. 257

Given the fact that the "room for error" is smaller for HG deals, in some sense it is even more crucial to conduct asset-level analysis.

31 March 2006


Exhibit 233: Subordination Levels and Loss Coverage of Junior AAA Tranches

Deal Name WAR Expected Loss Rate Junior AAA Subordination Loss Coverage Ratio

CDO 3 AA 0.2228% 5.00% 22.44

CDO 5 AA- 0.6008% 7.00% 11.65

CDO 6 AA/AA- 0.4003% 6.00% 14.99

CDO 7 AA/AA- 0.2773% 6.00% 21.64

CDO 8 AA/AA- 0.3918% 10.00% 25.52

CDO 11 AA- 0.2620% 5.00% 19.09

CDO 12 AA-/A+ 0.6328% 10.00% 15.80

CDO 13 AA/AA- 0.4045% 6.00% 14.83

CDO 14 AA- 0.6112% 7.00% 11.45

CDO 15 AA/AA- 0.3771% 8.00% 21.21

CDO 16 A+ 0.8035% 9.00% 11.20

CDO 18 AA/AA- 0.6505% 9.00% 13.84

CDO 19 AA 0.4203% 6.00% 14.27

CDO 20 AA- 0.4522% 6.00% 13.27

CDO 21 A+ 0.7245% 12.00% 16.56

CDO 22 A+ 0.7522% 10.00% 13.29

CDO 23 AA/AA- 0.4083% 7.00% 17.14

CDO 25 AA+ 0.3955% 5.00% 12.64

Source: Credit Suisse, Intex, Bloomberg

Exhibit 234: Break-even Default Rate vs. Implied Annual Default Rate

Deal Name

Implied Annual

Default Rate*

Break-even

Default Rate(20% CPR)

Break-even

Default Rate(30% CPR)

Ratio

(20% CPR)Ratio

(30% CPR)

CDO 3 0.10% 7.40% 8.50% 74.74 85.85

CDO 5 0.27% 6.80% 7.90% 25.47 29.59

CDO 6 0.18% 5.20% 5.80% 29.23 32.60

CDO 7 0.12% 4.60% 5.10% 37.33 41.38

CDO 8 0.17% 8.20% 10.00% 47.09 57.43

CDO 11 0.12% 3.40% 3.60% 29.20 30.92

CDO 12 0.28% 9.40% 12.40% 33.42 44.09

CDO 13 0.18% 4.30% 5.00% 23.92 27.82

CDO 14 0.27% 8.50% 10.10% 31.29 37.18

CDO 15 0.17% 4.70% 5.40% 28.04 32.22

CDO 16 0.36% 6.30% 7.30% 17.64 20.44

CDO 18 0.29% 5.80% 6.90% 20.06 23.87

CDO 19 0.19% 4.20% 4.90% 22.48 26.23

CDO 20 0.20% 5.70% 6.20% 28.36 30.85

CDO 21 0.32% 7.80% 8.90% 24.22 27.64

CDO 22 0.33% 5.70% 6.70% 17.05 20.04

CDO 23 0.18% 4.30% 4.80% 23.69 26.45

CDO 25 0.18% 7.60% 8.10% 43.24 46.08


* This is the annual default rate implied from the expected loss rates calculated by using a recovery rate of 55% and dividing by 5 years.


STRUCTURED PRODUCTS RESEARCH

Gail Lee, Managing Director

Global Head of Structured Products Research

+1 212 325 1214

Bunt Ghosh, Managing Director

Global Head of Fixed Income Research

+44 20 7888 3042

NORTH AMERICA Eleven Madison Avenue, New York, NY 10010

Asset-Backed Securities (ABS)

Rod Dubitsky, Managing Director Rajat Bhu, Vice President Chris Fenske, Vice President Jay Guo, Vice President

Senior Strategist, Group Head +1 212 325 4740 [email protected]

+1 212 325 5410 [email protected]



Shumin Li, Vice President Lidia Dumitrascu, Associate Larry Yang, Associate Christopher Mellia, Analyst

+1 212 325 2957

[email protected]




Collateralized Debt Obligations (CDO)

David Yan, Vice President Stephen Chow, Associate Neil Desai, Analyst Willie Green





Commercial Mortgage Backed Securities (CMBS)

Gail Lee, Managing Director Paul Fitzsimmons, Vice President Manish Rajguru, Vice President Serif Ustun, Vice President





Mortgage Backed Securities — Residential (MBS)

Satish Mansukhani, Managing Director Mahesh Swaminathan, Director Adama Kah, Vice President Chandrajit Bhattacharya, Vice President





Sergei Ivanov, Vice President Mutaz Qubbaj, Associate



EUROPE – Structured Products (All) One Cabot Square, London E14 4QJ, United Kingdom

Recai Güneşdoğdu, Director Tim Francis, Associate Michael Tian, Associate European Head +44 20 7883 7978 [email protected]



JAPAN – Structured Products (All) Izumi Garden Tower, 1-6 Roppongi 1-Chome, Minato-ku, Tokyo 106-6024

Kenji Toukaku, Director Kaoru Kondo, Associate Japan Head + 81 3 4550 7172 [email protected]

[email protected] +81 3 4550 7171

For general inquiries or to be added to a distribution list, please contact:

Angela Chuang ([email protected]) or Werner Pauliks ([email protected])


Disclosure Appendix

Analyst Certification David Yan and Stephen Chow each certify, with respect to the companies or securities that he or she analyzes, that (1) the views expressed in this report accurately reflect his or her personal views about all of the subject companies and securities and (2) no part of his or her compensation was, is or will be directly or indirectly related to the specific recommendations or views expressed in this report.

Important Disclosures Credit Suisse's policy is only to publish investment research that is impartial, independent, clear, fair and not misleading. For more detail, please refer to Credit Suisse's Policies for Managing Conflicts of Interest in connection with Investment Research: http://www.csfb.com/research-and-analytics/disclaimer/managing_conflicts_disclaimer.html Credit Suisse’s policy is to publish research reports as it deems appropriate, based on developments with the subject issuer, the sector or the market that may have a material impact on the research views or opinions stated herein. The analyst(s) involved in the preparation of this research report received compensation that is based upon various factors, including Credit Suisse's total revenues, a portion of which are generated by Credit Suisse's Investment Banking and Fixed Income Divisions. Credit Suisse may trade as principal in the securities or derivatives of the issuers that are the subject of this report. At any point in time, Credit Suisse is likely to have significant holdings in the securities mentioned in this report. As at the date of this report, Credit Suisse acts as a market maker or liquidity provider in the debt securities of the subject issuer(s) mentioned in this report. For important disclosure information on securities recommended in this report, please call +1-212-538-7625. For the history of any relative value trade ideas suggested by the Fixed Income research department over the previous 12 months, please view the document at http://research-and-analytics.csfb.com/docpopup.asp?docid=35321113&type=pdf. Credit Suisse clients with access to the Locus website may refer to http://www.credit-suisse.com/locus. For the history of recommendations provided by Technical Analysis, please visit the website at http://www.credit-suisse.com/techanalysis. Credit Suisse does not provide any tax advice. Any statement herein regarding any US federal tax is not intended or written to be used, and cannot be used, by any taxpayer for the purposes of avoiding any penalties.

Emerging Markets Bond Recommendation Definitions Buy: Indicates a recommended buy on our expectation that the issue will deliver a return higher than the risk-free rate. Sell: Indicates a recommended sell on our expectation that the issue will deliver a return lower than the risk-free rate.

Corporate Bond Fundamental Recommendation Definitions Buy: Indicates a recommended buy on our expectation that the issue will be a top performer in its sector. Outperform: Indicates an above-average total return performer within its sector. Bonds in this category have stable or improving credit profiles and are undervalued, or they may be weaker credits that, we believe, are cheap relative to the sector and are expected to outperform on a total-return basis. These bonds may possess price risk in a volatile environment. Market Perform: Indicates a bond that is expected to return average performance in its sector. Underperform: Indicates a below-average total-return performer within its sector. Bonds in this category have weak or worsening credit trends, or they may be stable credits that, we believe, are overvalued or rich relative to the sector. Sell: Indicates a recommended sell on the expectation that the issue will be among the poor performers in its sector. Restricted: In certain circumstances, Credit Suisse policy and/or applicable law and regulations preclude certain types of communications, including an investment recommendation, during the course of Credit Suisse's engagement in an investment banking transaction and in certain other circumstances.

Corporate Bond Risk Category Definitions In addition to the recommendation, each issue may have a risk category indicating that it is an appropriate holding for an "average" high yield investor, designated as Market, or that it has a higher or lower risk profile, designated as Speculative and Conservative, respectively.

Credit Suisse Credit Rating Definitions Credit Suisse assigns rating opinions to investment-grade and crossover issuers. Ratings are based on our assessment of a company's creditworthiness and are not recommendations to buy or sell a security. The ratings scale (AAA, AA, A, BBB, BB, B) is dependent on our assessment of an issuer's ability to meet its financial commitments in a timely manner. Within each category, creditworthiness is further detailed with a scale of High, Mid, or Low – with High being the strongest sub-category rating: High AAA, Mid AAA, Low AAA – obligor's capacity to meet its financial commitments is extremely strong; High AA, Mid AA, Low AA – obligor's capacity to meet its financial commitments is very strong; High A, Mid A, Low A – obligor's capacity to meet its financial commitments is strong; High BBB, Mid BBB, Low BBB – obligor's capacity to meet its financial commitments is adequate, but adverse economic/operating/financial circumstances are more likely to lead to a weakened capacity to meet its obligations; High BB, Mid BB, Low BB – obligations have speculative characteristics and are subject to substantial credit risk; High B, Mid B, Low B – obligor's capacity to meet financial commitments is very weak and highly vulnerable to adverse economic, operating, and financial circumstances. Credit Suisse's rating opinions do not necessarily correlate with those of the rating agencies.

References in this report to Credit Suisse include all of the subsidiaries and affiliates of Credit Suisse operating under its investment banking division. For more information on our structure, please use the following link: http://www.credit-suisse.com/en/who_we_are/ourstructure.html. This report is not directed to, or intended for distribution to or use by, any person or entity who is a citizen or resident of or located in any locality, state, country or other jurisdiction where such distribution, publication, availability or use would be contrary to law or regulation or which would subject Credit Suisse or its affiliates (“CS”) to any registration or licensing requirement within such jurisdiction. All material presented in this report, unless specifically indicated otherwise, is under copyright to CS. None of the material, nor its content, nor any copy of it, may be altered in any way, transmitted to, copied or distributed to any other party, without the prior express written permission of CS. All trademarks, service marks and logos used in this report are trademarks or service marks or registered trademarks or service marks of CS or its affiliates. The information, tools and material presented in this report are provided to you for information purposes only and are not to be used or considered as an offer or the solicitation of an offer to sell or to buy or subscribe for securities or other financial instruments. CS may not have taken any steps to ensure that the securities referred to in this report are suitable for any particular investor. CS will not treat recipients of this report as its customers by virtue of their receiving this report. The investments and services contained or referred to in this report may not be suitable for you and it is recommended that you consult an independent investment advisor if you are in doubt about such investments or investment services. Nothing in this report constitutes investment, legal, accounting or tax advice, or a representation that any investment or strategy is suitable or appropriate to your individual circumstances, or otherwise constitutes a personal recommendation to you. CS does not advise on the tax consequences of investments and you are advised to contact an independent tax adviser. Please note in particular that the bases and levels of taxation may change. Information and opinions presented in this report have been obtained or derived from sources believed by CS to be reliable, but CS makes no representation as to their accuracy or completeness. CS accepts no liability for loss arising from the use of the material presented in this report, except that this exclusion of liability does not apply to the extent that such liability arises under specific statutes or regulations applicable to CS. This report is not to be relied upon in substitution for the exercise of independent judgment. CS may have issued, and may in the future issue, other reports that are inconsistent with, and reach different conclusions from, the information presented in this report. Those reports reflect the different assumptions, views and analytical methods of the analysts who prepared them and CS is under no obligation to ensure that such other reports are brought to the attention of any recipient of this report. CS may, to the extent permitted by law, participate or invest in financing transactions with the issuer(s) of the securities referred to in this report, perform services for or solicit business from such issuers, and/or have a position or holding, or other material interest, or effect transactions, in such securities or options thereon, or other investments related thereto. In addition, it may make markets in the securities mentioned in the material presented in this report. CS may have, within the last three years, served as manager or co-manager of a public offering of securities for, or currently may make a primary market in issues of, any or all of the entities mentioned in this report or may be providing, or have provided within the previous 12 months, significant advice or investment services in relation to the investment concerned or a related investment. Additional information is, subject to duties of confidentiality, available on request. Some investments referred to in this report will be offered solely by a single entity and in the case of some investments solely by CS, or an associate of CS or CS may be the only market maker in such investments. Past performance should not be taken as an indication or guarantee of future performance, and no representation or warranty, express or implied, is made regarding future performance. Information, opinions and estimates contained in this report reflect a judgement at its original date of publication by CS and are subject to change without notice. The price, value of and income from any of the securities or financial instruments mentioned in this report can fall as well as rise. The value of securities and financial instruments is subject to exchange rate fluctuation that may have a positive or adverse effect on the price or income of such securities or financial instruments. Investors in securities such as ADR’s, the values of which are influenced by currency volatility, effectively assume this risk. Structured securities are complex instruments, typically involve a high degree of risk and are intended for sale only to sophisticated investors who are capable of understanding and assuming the risks involved. The market value of any structured security may be affected by changes in economic, financial and political factors (including, but not limited to, spot and forward interest and exchange rates), time to maturity, market conditions and volatility, and the credit quality of any issuer or reference issuer. Any investor interested in purchasing a structured product should conduct their own investigation and analysis of the product and consult with their own professional advisers as to the risks involved in making such a purchase. Some investments discussed in this report may have a high level of volatility. High volatility investments may experience sudden and large falls in their value causing losses when that investment is realised. Those losses may equal your original investment. Indeed, in the case of some investments the potential losses may exceed the amount of initial investment and, in such circumstances, you may be required to pay more money to support those losses. Income yields from investments may fluctuate and, in consequence, initial capital paid to make the investment may be used as part of that income yield. Some investments may not be readily realisable and it may be difficult to sell or realise those investments, similarly it may prove difficult for you to obtain reliable information about the value, or risks, to which such an investment is exposed. This report may provide the addresses of, or contain hyperlinks to, websites. Except to the extent to which the report refers to website material of CS, CS has not reviewed any such site and takes no responsibility for the content contained therein. Such address or hyperlink (including addresses or hyperlinks to CS’s own website material) is provided solely for your convenience and information and the content of any such website does not in any way form part of this document. Accessing such website or following such link through this report or CS’s website shall be at your own risk. This report is issued and distributed in Europe (except Switzerland) by Credit Suisse Securities (Europe) Limited, One Cabot Square, London E14 4QJ, England, which is regulated in the United Kingdom by The Financial Services Authority (“FSA”). This report is being distributed in Germany by Credit Suisse Securities (Europe) Limited Niederlassung Frankfurt am Main regulated by the Bundesanstalt fuer Finanzdienstleistungsaufsicht ("BaFin"). This report is being distributed in the United States and Canada by Credit Suisse Securities (USA) LLC; in Switzerland by Credit Suisse; in Brazil by Banco de Investimentos Credit Suisse (Brasil) S.A; in Japan by Credit Suisse First Boston Securities (Japan) Limited; elsewhere in Asia/ Pacific by whichever of the following is the appropriately authorised entity in the relevant jurisdiction: Credit Suisse (Hong Kong) Limited, Credit Suisse Equities (Australia) Limited, Credit Suisse Securities (Thailand) Limited, Credit Suisse Securities (Malaysia) Sdn Bhd, Credit Suisse Singapore Branch, and elsewhere in the world by the relevant authorised affiliate of the above. Research on Taiwanese securities produced by Credit Suisse, Taipei Branch has been prepared by a registered Senior Business Person. Research provided to residents of Malaysia is authorised by the Head of Research for Credit Suisse Securities (Malaysia) Sdn Bhd, to whom they should direct any queries on +603 2723 2020. This research may not conform to Canadian disclosure requirements. In jurisdictions where CS is not already registered or licensed to trade in securities, transactions will only be effected in accordance with applicable securities legislation, which will vary from jurisdiction to jurisdiction and may require that the trade be made in accordance with applicable exemptions from registration or licensing requirements. Non-U.S. customers wishing to effect a transaction should contact a CS entity in their local jurisdiction unless governing law permits otherwise. U.S. customers wishing to effect a transaction should do so only by contacting a representative at Credit Suisse Securities (USA) LLC in the U.S. Please note that this research was originally prepared and issued by CS for distribution to their market professional and institutional investor customers. Recipients who are not market professional or institutional investor customers of CS should seek the advice of their independent financial advisor prior to taking any investment decision based on this report or for any necessary explanation of its contents. This research may relate to investments or services of a person outside of the UK or to other matters which are not regulated by the FSA or in respect of which the protections of the FSA for private customers and/or the UK compensation scheme may not be available, and further details as to where this may be the case are available upon request in respect of this report. Copyright © 2006 CREDIT SUISSE GROUP and/or its affiliates. All rights reserved.

CDO Handbook Final

Documents

Transcript of CDO Handbook Final