Deal Sponsor and Credit Risk of ABS and MBS Securities

Jian Hu
Director of Structured Finance Default Research Credit Policy Research

December 6, 2006

Presentation Outline
Introduction Distribution of deals by sponsor type Variation in Credit Performance by Sponsor Type
In the aggregate Across asset classes Within asset class and by tranche rating

Variations in Spreads by Sponsor Type

Aaa spreads by asset class Baa spreads by asset class

Recap of Main Findings and Possible Reasons

Market Regularly Studies Role of Sponsor

When GMs downgrade was in the news in 2005 Auto spreads weakened a touch more this week with
investors being more cautious. GMAC Auto ABS spreads widened further, contributing to increased tiering between GMAC and other names. Spread on Ford ABS bonds, for example, have held in fairly well.*

After the downgrade of Discover Bank in 2005

With the contraction in Credit Card ABS supply, due in part to portfolio acquisition trends, and spreads among issuers as compressed as they are, Discover bonds traded merely a hair wider on the news of the spin off.*

New ABS regulation (2004) requires the reporting and disclosure of detailed sponsor information
* JP Morgan Global ABS/CDO Weekly Market Snapshot, 2005 3

Findings from Moodys Default/Loss Studies

Poor performance in some ABS asset classes were related to a few deal sponsor failures:
Heilig-Meyers (servicing/operational failure) DVI, NCFE (fraud) Green Tree/Conseco (Market meltdown/servicing)

More than 40% of the defaulted HEL bonds were associated with one originator/sponsor: Quality Mortgage/DLJ, and were issued before 1997

How Can Sponsor Affect Deal Performance?

Four possible ways:
Origination/asset selection criteria Servicing quality Potential debt consolidation in bankruptcy (rare) Supporting for troubled deals (has become rare)

Method of Analysis in This Study

Sample consists of all US ABS, RMBS, CMBS tranches issued since 1993 Sponsors classified by rating class and industry Credit performance metrics
Lifetime downgrade rate Lifetime impairment rate Coupon spread at issuance

Compare performance by sponsor type

across and within asset classes With and without controlling for tranche rating
6

Main Findings
The distribution of sponsor type within an asset class had a stronger impact on security performance than the nature of a deals own particular sponsor For some ABS asset classes such as credit card, small business loan, and equipment lease ABS, sponsor type has a strong impact on performance and spread. For most other asset classes especially those backed by mortgage loans, the impact was weaker or non-existent After controlling for a sponsors rating, tranches sponsored by banks performed generally better than those sponsored by specialty finance firms
7

Definition and Data Sample Distribution

Definition of Deal Sponsor

Sponsor is either the seller, or if the seller is a wholly-owned subsidiary, the sponsor is the parent of the seller More formally, according to SEC Reg AB,
sponsor is an entity that organizes and initiates a structured finance transaction by selling or transferring assets, either directly or indirectly, including through an affiliate, to the SPV

Classification of Sponsors
Two broad rating categories
Investment grade Non-investment-grade (either speculative grade or unrated)

Four industry categories

US banks, insurance companies, thrifts
Specialty finance companies, mortgage banks, REIT, others Securities firms, foreign banks, investment firms Industrial captive finance companies

10

Share of Deals by Sponsor Types

By Sponsor Rating
NR or NA 31% Aa or above 27%

By Sponsor Industry
Unknown 6% Captive 14%

Banks 28%

SG 6% Baa 6% Single-A 30%

Securities 25% Specialty 27%

For all 8,914 deals issued during 1993-2006H1

11

Share of Deals by Sponsor Rating Within Industries

IG Non-IG 100% 80% 60% 40% 20% 0%
Banks Specialty Securities Captive

12

Share of Deals by Sponsor Type

By Sponsor Rating
Non-IG IG 100% 80% 60% 40% 20% 0%
Student MH Specialized RMBS Cards Autos CMBS HEL
100% 80% 60% 40% 20% Specialized 0% Cards Autos Student MH RMBS CMBS

By Sponsor Industry
Captive Securities Specialty Banks

HEL

13

Distribution of Deals over Time in RMBS & HEL

RMBS
Captive Securities Specialty Banks 100% 80% 60% 40% 20% 0%
1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006

HEL
Captive Securities Specialty Banks 100% 80% 60% 40% 20% 0% 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006
14

Compare Performance by Sponsor Type

15

Analyze Sponsor Effects on Credit Performance

In the aggregate
Sponsor industry effect controlling for sponsor rating effect Sponsor rating effect controlling for sponsor industry effect

Comparison across asset classes Comparison within asset classes and by tranche rating Comparison of downgrade severity within asset class Analysis based on securities issued during 19932002
16

Lifetime Downgrade Rates Vary by Sponsor Type

25%

20%

15%

10%

5%

0% IG Non-IG Banks Captive Securities Specialty

17

Lifetime Impairment Rates Also Vary by Sponsor Type

12%

9%

6%

3%

0% IG Non-IG Banks Captive Securities Specialty

18

Sponsor Industry Effect Is Evident Within Rating Classes

Lifetime Downgrade Rate Lifetime Impairment Rate

25% 20% 15% 10% 5% 0%

Banks Captive Securities-DLJ Specialty

12% 9% 6% 3% 0%

Banks Captive Securities-DLJ Specialty

IG

Non-IG

IG

Non-IG

19

Sponsor Rating Effect Is Also Evident Within Industries

Lifetime Downgrade Rate Lifetime Impairment Rate

25% 20% 15% 10% 5% 0%

IG

Non-IG

12% 9% 6% 3% 0%

IG

Non-IG

Banks

Banks

Captive

Specialty

Securitie s-DLJ

Securitie s-DLJ

Captive

Specialty
20

DLJ/Quality Deals Had A Big Impact in HEL

DLJ deals will be excluded from within-asset-class comparisons
IG ex DLJ 20% 15% 10% 5% 0%
Lifetime Impairment Rate Lifetime Downgrade Rate

Non-IG

IG incl DLJ

21

100%

10%

20%

30%

40%

50%

60%

70%

80%

90%

0% HELOC Student Timeshr Utility PrimeAuto Ag&Ind OtherRMBS Floorplan Consumer BankCard RMBS NIM RetailCard OtherABS Receiv Lease HLTV ResecRMBS SubAuto Transpt CMBS HEL SmBusn Euip HIL ManuH MutuFund FrLn Aircraft Healthcare

Some ABS asset classes performed poorly

Lifetime downgrade rate Lifetime Impairment Rate

Credit Performance Vary by Asset Class

22

Asset Class Downgrade Rate Related to Share of Non-IG Sponsors

Downgrade rate higher if share of non-IG sponsors is higher
100% 90% 80%
Lifetime Downgrade Rate
Healthcare

Regression Line: y = 0.9016x - 0.3584x + 0.0563 R2 = 0.3882

Aircraft Lease Franchise Loan Mutual Fund Fee Manu. Housing

70% 60% 50% 40% 30% 20% 10% 0% 0% 10% 20% 30% 40%
CMBS Consum Ln HEL RMBS Bank Card Retail Card

Equip. Lease Small Business Transport HLTV Lease

Home Improve

Subprime Auto

50%

60%

70%

80%

90%

100%

Share of Tranches by Non-IG Sponsors

23

Impairment Rates Exhibit Similar Pattern

Impairment rate higher if share of non-IG sponsors is higher
100% 90% 80% Lifetime Impairment Rate 70% 60% 50% 40% 30% 20% 10% 0% 0% 10%
CMBS Bank RMBS Card Retail Card HEL HLTV Lease Aircraft Lease Franchise Loan Manu. H Mutual Fund Fee

Regression Line: y = 0.9597x 2 - 0.588x + 0.0797 R2 = 0.3955

Healthcare

Equip. Lease Subprime Small Auto Business

20%

30%

40%

50%

60%

70%

80%

90%

100%

Share of Tranches by Non-IG Sponsors

24

Asset Class Downgrade Rate Also Related to Share of Specialty Firms

Downgrade rate higher if share of specialty firms is higher
100% 90% 80% Lifetime Downgrade Rate 70%
Mutual Fund

Regression Line: y = 1.4511x2 - 0.7362x + 0.1008 R2 = 0.6121

Aircraft Lease

Healthcare

Fran. Loan

60% 50% 40% 30% 20% 10% 0% 0% 10% 20% 30% 40% 50% 60%
Bank Card Prime CMBS Auto Transp RMBS HLTV Consume HEL Lease Equip. Lease

Manu. H

Small Busn

Home Improve Subprime Auto Student

70%

80%

90%

100%

Share of Tranches Sponsored by Specialty Finance Companies

25

Sponsor Rating Impact Evident within Asset Class

IG-sponsored tranches performed better in all but two classes

80% 70% 60% 50% 40% 30% 20% 10% Small Business Loan Subprime Auto Specialized ABS Manu. Housing 0% Student Loan Retail Card Prime Auto Bank Card RMBS CMBS HEL Equip. Lease

Lifetime Downgrade Rates

IG Non-IG

26

Strong Sponsor Rating Impact on Downgrade Risk of Highly Rated Tranches

Aaa tranches were affected the most
Logistic regressions run for each asset class with a sponsor rating dummy, as well as tranche rating dummies. Downgrade is the dependent variable. Tranche Rating Aaa Aa A Baa SG All Assets Auto 9.62 -0.30 11.65 10.65 -0.79 Card Special ABS MH 0.06 -0.46 -1.42 -1.51 14.05 HEL ex. DLJ 8.68 -0.11 0.24 RMBS CMBS -13.20 -13.56 -14.28 -0.36 -1.13

1.55 0.64 1.20 0.61

0.15

4.13
11.41

2.02 0.90 0.98

0.53 -0.79

2.86
-0.57 0.05 0.34

2.61 2.22
12.69

0.37
-0.54

0.92

Large bold-faced values (circled) are significantly positive 27

Sponsor Rating Impact on Impairment Risk Weaker

Positive coefficients for most ABS rating categories above Baa
Logistic regressions run for each asset class with a sponsor rating dummy, as well as tranche rating dummies. Impairment is the dependent variable. Tranche Rating Aaa Aa A Baa SG All Assets Auto Card Special ABS 10.49 MH HEL ex. DLJ -0.37 -0.53 0.54 RMBS CMBS

1.21 1.10 1.46 0.77 0.58

-0.13 -0.39 10.97 8.88 -1.67

10.72 11.15 10.87 11.59 10.48

-1.55 0.00 -0.54 -1.11 13.87

-0.55 -11.25 0.49 0.12

0.22 0.14 -11.66 -13.03 -0.72

1.46
0.72 0.49 -0.15

0.61
-1.36

0.92

Large bold-faced values (circled) are significantly positive 28

Minimal Specialty Company Effect within Asset Classes

Fewer cells with positive coefficients
Logistic regressions run for each asset class with a specialty finance dummy, as well as tranche rating dummies. Impairment is the dependent variable. Tranche Rating Aaa Aa A Baa SG All Assets Auto Card Special ABS MH HEL ex. DLJ RMBS CMBS

1.56 1.69 2.09 1.66 0.73

-0.11 -1.22 11.78 8.36 -1.55

-8.02 -10.90 -8.01 -10.78 na

1.83 1.63 1.93 1.97

0.66

-2.29 -0.02 0.95 -0.72 0.45

-0.26 -0.34 0.95

-0.37 -10.65 -13.01 -0.97 -0.34

0.15 0.11 -9.79 -11.01 -0.60

0.79
0.15

Large bold-faced values (circled) are significantly positive 29

Number of Notches Downgraded 10 12 14 16 0 CMBS 2 4 6 8

IG

RMBS HEL High LTV Bank Credit Card Aircraft Lease Mutual Fund Fees Franchise Loans MH Subprime Auto Equipment Lease

Downgrade Severity Varied by Sponsor Rating

Large differences in ABS; Opposite effect in CMBS & RMBS

Non-IG

30

Comparing Spreads by Sponsor Type

31

Analyze Spreads by Sponsor Type

Spread analysis uses a more recent portion of our data sample from 1998 to 2006H1* Spread analysis reinforces, or in some cases strengthens, our prior results from performance comparisons Spread is more complex to analyze because of interest rate type, maturity, liquidity, and prepayment effects

Note: This data sample updates the one that was used for Moodys Special Comment, The Relationship Between Par Coupon Spreads and Credit Ratings in US Structured Finance, December 2005 32

Coupon Spreads Vary by Sponsor Rating

40 35 30 25 20 15 10 5 0

IG

Non-IG

Basis Points

ABS

HEL

RMBS

CMBS

33

basis points 10 20 30 40 50 60 70 0
AutoPrime Floorplan BankCard AgriInd ABSoth Utility Student RetailCard ManuH Heloc Lease Aircraft Euip HEL HLTV CMBS Receiv OtherRMBS RMBS SmBusn ResecRMBS Transpt Timeshr AutoSub FrLn Health
34

Wide Aaa Spread Variation Across Asset Classes

Median Spread & Non-IG Sponsors by Asset Class

Relationship is weaker than that based on performance
70 60 50 40 30 20 10 0 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Share of Tranches by Non-IG Sponsors
35
Consum Ln RMBS Resec RMBS CMBS HELOC Utility Other RMBS HLTV Lease Retail Card Ag & Ind EQuip Flplan Prime Auto Timeshr

Regression Line: y = 18.885x + 19.073 R2 = 0.1441

Median Aaa Floating Rate Spread

Healthcare Fran Ln Subprime Auto Transpt Small Busn Receiv Equip Lease Manu H Student Home Improve

Bank Card

Airc Lease Other ABS

100

120

140

160

20

40

60

80

Prime Auto Fixed Floorplan Floating Student Floating Bank Credit Card Floating Prime Auto Floating Retail Credit Card Floating HEL Floating Rate

IG

High LTV Floating Subprime Auto Fixed Equipment Lease Floating RMBS Floating CMBS Floating CMBS Fixed Small Business Loans Floating HEL Fixed Rate MH Fixed High LTV Fixed Franchise Loan Fixed RMBS Fixed

Sponsor Rating Impact on Spreads by Asset Class

Spreads were lower in IG for 15 of the 19 asset classes

Aaa Tranche Spread

Non-IG

36

100

150

200

250

300

350

400

450

50

Prime Auto Fixed Retail Credit Card Floating CMBS Fixed RMBS Floating Bank Credit Card Floating Small Business Loans Floating RMBS Fixed

IG

Non-IG

High LTV Floating CMBS Floating HEL Fixed Rate HEL Floating Rate MH Fixed High LTV Fixed NIM Fixed

Sponsor Rating Also Affects Baa Spreads

Spreads were lower in IG for 12 of the 14 asset classes

Baa Tranche Spread

37

Recap of Main Findings

The distribution of sponsor type within an asset class had a stronger impact on security performance than the nature of a deals own particular sponsor For some ABS asset classes such as credit card, small business loan, and equipment lease ABS, sponsor type has a strong impact on performance and spread. For most other asset classes especially those backed by mortgage loans, the impact was weaker or non-existent After controlling for a sponsors rating, tranches sponsored by banks performed generally better than those sponsored by specialty finance firms
38

Possible Reasons for the Findings

Collateral: Competitive forces may cause lower rated
sponsors and specialty finance companies to be most active in market segments where collateral is risky and pool performance is hard to predict

Servicing: Deal performance may be inversely related to

the credit quality of the sponsor since the pools underlying obligors may be less likely to service their debts if the servicer is bankrupt

Agency problems: may be less severe for more highly

rated sponsors if their expected long-term participation in securitization makes them more likely to be consistent in their underwriting standards, selection of assets, and servicing
39

Update on Loss Given Default for Structured Finance

Julia Tung Vice President

December 6, 2006

Agenda
Measuring LGD Key Findings from LGD Data LGD Estimation Methods

41

Definition of Loss Severity Rate Given Default (LGD)

LGD is defined as the sum of the discounted present values of net interest shortfalls and net principal losses each period The securitys coupon rate is used as the discount rate LGD is expressed as a share of the securitys principal balance at origination date, default date, or cohort date

LGDk ,t

IS s + LPs (1 + c ) s k +1 s = s =k Bk
t

where ISs denotes interest shortfall, LPs denotes loss of principal, cs is the discount rate, and Bk is the outstanding principal balance.
42

LGD by original balance vs. default balance

Difference due to:
Dividing by original balance vs. default balance (principal payments prior to default) Discounting to origination vs. default date

LGD by original balance will always be less than or equal to LGD by default balance

43

Difficulties in Measuring LGD

Trading prices are generally not available Losses accrue gradually over time so final losses are not known for most defaulted securities
LGD is known for Matured Defaults (securities with zero outstanding balance) LGD needs to be estimated for Non-Matured Defaults (securities with positive principal balances)

Different servicers/trustees use different definitions and formats in their reports Actual payment/loss data are sometimes incomplete and inaccurate
44

Data Sample of Uncured Payment Defaults (with loss info) 1993-2006H1

1000 900 800 700 600 500 400 300 200 100 0 All Matured Non-Matured US ABS Global CDOs US RMBS/HEL US CMBS

Total of 925 payment defaults US ABS has largest share of payment defaults 39% of total sample of payment defaults have matured
45

Key Findings from LGD Data

46

Key Findings
Principal write-downs are a much larger source of losses than missed interest Missed interest tends to be more important for non-matured defaults than matured defaults LGD on matured defaults overestimate expected LGDs on non-matured defaults LGD is a decreasing function of tranche size and time to default LGD varies by asset class LGD is only weakly related to ratings at origination
47

Principal write-downs are a much larger source of losses than missed interest
Average final loss severity rates for matured defaults
80% 70% 60% 50% 40% 30% 20% 10% 0% Total SF US RMBS/HEL US CMBS Global CDOs US ABS

Principal Loss Interest Shortfall

Losses measured as a share of the tranches original balance.

48

Missed interest tends to be more important for nonmatured defaults than matured defaults
Average loss severity to date for non-matured defaults
25%
Principal Loss Interest Shortfall

20%

15%

10%

5%

0% Total SF US CMBS US RMBS/HEL US ABS Global CDOs

Losses measured as a share of the tranches original balance.

49

LGD on matured defaults overestimate expected LGDs on non-matured defaults

Average estimated LGD for non-matured defaults ~80% of average LGD for matured defaults
100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% US RMBS/HEL US MH Matured LGD Non-Matured Estimated LGD

Losses measured as a share of the tranches default balance.

50

Differences between Matured and Non-Matured Defaults

Default balance Months to default Original rating

US RMBS/HEL Matured Non-Matured 81.6% 79.4% 45.1 54.4 Ba1 Baa3

US MH Matured Non-Matured 98.9% 98.7% 44.1 68.8 Baa2 Baa1

Matured defaults had higher balances at default, defaulted more quickly, and had lower original ratings These indicate that the matured sample had lower credit quality and more junior tranches leading to higher LGD relative to the non-matured sample

51

Estimated Final LGD (by original balance) for the Combined Sample
Average LGD: 50.9% Median LGD: 57.1%
14% Percentage of Defaults 12% 10% 8% 6% 4% 2% 0% 5% 10% 15% 20% 25% 30% 35% 40% 45% 50% 55% 60% 65% 70% 75% 80% 85% 90% 95% 100%

LGD (% original balance)

52

Estimated Final LGD (by default balance) for the Combined Sample
Average LGD: 71.8% Median LGD: 89.1%
40% 35% Percentage of Defaults 30% 25% 20% 15% 10% 5% 0% 5% 10% 15% 20% 25% 30% 35% 40% 45% 50% 55% 60% 65% 70% 75% 80% 85% 90% 95% 100%

LGD (% default balance)

53

LGD is a decreasing function of tranche size

90% 80% 70% 60% 50% 40% 30% 20% 10% 0% <=2.5% 2.5%-5% 5%-10% 10%-20% >20% Tranche Size (% deal balance)

by original balance by default balance

54

LGD is a decreasing function of time to default

90% 80% 70% 60% 50% 40% 30% 20% 10% 0% <=3 yrs 3-4 yrs 4-5 yrs Time to Default 5-6 yrs > 6 yrs

by original balance by default balance

55

LGD varies by sector

100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% Total SF US RMBS/HEL US CMBS US ABS Global CDOs by original balance by default balance

56

Reasons for Sectoral Differences

Amortization rates
RMBS/HEL securities amortize more quickly Avg balance at default :
RMBS/HEL 80.9% ABS 98.5% CDOs 97.4%

Structure
Interest subordination among MH and PIKing among CDOs

Credit cycle
Strong US housing market versus extreme downturn for MH and HY CBOs

57

LGD is only weakly related to ratings at origination

LGD rank-ordered by original rating for RMBS/HEL, but not for ABS and CDOs
60% 50% 40% 30% 20% 10% 0% Aaa Aa A Baa Ba B Original Rating US RMBS/HEL
90% 80% 70% 60% 50% 40% 30% 20% 10% 0% Aa A Baa Original Rating Ba B US ABS Global CDOs

Losses measured as a share of the tranches original balance.

58

LGD Estimation Methods

59

Is it worth the trouble to produce a refined LGD estimate?

Estimated LGD for non-matured defaults
LGD to date 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% US RMBS/HEL US MH Global CDOs Estimated Final LGD Maximum Possible LGD

LGD is measured as a share of the tranches default balance.

60

Relationship between Maximum and Predicted Incremental Loss

100% 80% Predicted Incremental Loss 60% 40% 20% 0% -20% -40% -60% Maximum Incremental Loss 0% 20% 40% 60% 80% 100%

Many tranches predicted to lose close to maximum possible (57% predicted to lose > 90% of max) But 35% predicted to lose 50% of max
Losses measured as a share of the tranches default balance. 61

Forecasting LGD for Non-Matured Defaults

Tranche-level model developed for US RMBS and HEL in April 2004 (Measuring Loss Severity Rates of
Defaulted Residential Mortgage-Backed Securities: A Methodology)

LGD estimation method for CDOs published in March 2005 (Default & Loss Rates of U.S. CDOs: 19932003) Pool-level model recently created for MH ABS

62

LGD Estimation Method for RMBS/HEL Model Tranche Losses Directly

Forecast future tranche losses based only on tranchelevel information Three variations
Static model: forecast losses based on static variables such as time to default, tranche size, etc. Dynamic model: forecast remaining losses based on the speed of loss accumulation to date Blended model: combination of static and dynamic

Pros: requires minimal information and relatively simple implementation Cons: requires large enough sample of matured defaults and may not produce consistent tranche LGD estimates within a deal
63

LGD Estimation Method for CDOs Estimate Pool Losses through Pool Attributes
Forecast losses of the underlying assets based on pool attributes and propagate losses to the security level CDO LGD projection model
Apply default and recovery rates to the current performing par amount in the portfolio based on WARF Estimate excess spread in the deal based on I/C and cost of fund estimates Use the above estimates to project ultimate loss rate for each impaired tranche

Pros: does not require a sample of matured defaults and incorporates deal structure Cons: more complicated to implement and requires a rating on the pool to forecast losses

64

LGD Estimation Method for MH ABS Estimate Pool Losses through Historical Data
Forecast losses of the underlying assets based on historical pool data and propagate losses to the security level MH LGD projection model
Model pool net loss rate and principal payment rate based on historical pool data Project future pool losses and interest/principal payments based on these models Each period, distribute interest, principal and loss to the tranches assuming a simple waterfall Sum the present value of tranche losses to calculate LGD

65

LGD Estimation Method for MH ABS Estimate Pool Losses through Historical Data (cont.)
Highlights of the method
Uses both pool loss and principal payment information rather than simply cumulative loss Multi-period forecast rather than single period Produces a point estimate for the tranche LGD which is appropriate for defaulted tranches, but not for non-defaulted tranches, where a distribution of losses should be estimated

Pros: does not require a sample of matured defaults or a pool rating and incorporates deal structure Cons: even more complex implementation and requires historical pool data

66

LGD Estimation Method for the Rest Extrapolate based on Sector, Rating, etc.
Project LGD for tranches with no estimates based on average LGD of tranches with estimates by sector, rating, etc. CMBS and non-standard RMBS/HEL use average RMBS/HEL LGD CDOs use average CDO LGD Non-MH ABS
Formerly used a combination of RMBS/HEL and CDO LGD In the future will also incorporate ABS LGD

67

Future work
Details of MH LGD estimation method forthcoming Plan to re-estimate RMBS/HEL LGD by forecasting pool losses

68

Defaults and Recovery Rates of Preferred Stocks Research in Progress

Presented by: Praveen Varma VP-Senior Credit Officer

December 6, 2006

Agenda
Descriptive statistics Defaults and default rates Recovery rates Summary and conclusions

70

Objectives and Descriptive Statistics

71

Objectives
Update the industrys only prior study (1998) on the credit performance of preferred stocks Why analyze preferred stocks separately?
Potential selection bias issuers with preferred stock issuers might have different default rates In addition to default/bankruptcy risk, preferred stock issuers can skip dividends without prompting an acceleration of other obligations Preferred LGD differs from bond & loan LGD because of its very junior status & possible dividend omissions
72

Data Sources and Definitions

Ratings, defaults and recovery data are extracted from Moodys Ratings Database and Default Risk Service database which contains
8,000 bond, loan, & preferred stock defaults 15,000 prices on defaulted instruments

73

Data Sources and Definitions

Treatment of cumulative vs. non-cumulative
Data slightly murky as to which issue is cumulative and which is not Data clean up continues but not finished For simplicity, in this presentation, ordinary cumulative and non-cumulative preferred stocks are not treated as separate classes

Treatment of trust preferred and other hybrids

Analyzed separately from traditional preferred
74

Data Sources and Definitions

To be included in the sample, issuer must have issued rated preferred stock! Preferred stock defaults occur when
Dividends are omitted/skipped/deferred Missed payments or distressed exchange on bonds or loans Bankruptcy

75

Number of rated preferred stock and trust preferred issues have grown significantly over the years
Growth in Moodys Rated Issues
700 600 500 400 300 200 100 0 1980 1990 1995 2000 2005 2006
Preferred Trust Preferred

Total outstanding amounts as of Jan 1, 2006 Preferred Stocks : Approx USD 91 billion Trust Preferred Stocks: Approx USD 130 billion
76

Most of these preferred stocks and trust preferred stocks have investment grade ratings

B 3% Ba 4%

Caa-C 5%

Caa-C 3%

Aa +Aaa 20%

Baa 27% A 45%

Baa 23%

Ba 8%

A 45%

B 5%

Rating Distribution of Preferred Stock Issues

Aa 12%

Rating distribution of Trust Preferred Stock Issues 77

As of Jan 2006, more than 60% of preferred stock issuers are either banks or other financial institutions
80% 70% 60% 50% 40% 30% 20% 10% 0%
Banking Non Bank Finance Coporates Public Utilities Insurance
78

Preferred

Trust Preferred

Most of the rated preferred stock issuers are located in the US and Canada, about 20% in Europe

90% 80% 70% 60% 50% 40% 30% 20% 10% 0% US and Canada EMEA Asia Pacific
Preferred
Trust Preferred

79

Defaults, Default Rates & Recoveries

80

268 Issuers of Preferred Stock Defaulted on Their Preferred Stock Obligations During the 1980-2005 Period
Default Event Distribution

60%

50%

40%

30%

20%

10%

0% Dividend Omission Chapter 11 Missed Bond Payments Distressed Exchange Other

81

What Happens to Dividend Omissions? Some Are Resolved Favorably, Many Unfavorably

45% 40% 35% 30% 25% 20% 15% 10% 5% 0% Further default Arrears cleared In arrears Acquired Called Converted

Dividend Omission Resolution

82

Industrials Have Disproportionate Share of Defaults

Industry Distribution of Defaults
90% 80% 70% 60% 50% 40% 30% 20% 10% 0% Industrial Banking, Finance and Insurance Public Utility Other nonTransportation bank, real estate finance, thrifts
83

Share of defaults Share of rated issuers

Default Rates
Default rates for preferred stock calculated in 2 ways 1. Including bond defaults, bankruptcies, and only dividend omissions that ultimate result in broader defaults 2. Including bond defaults, bankruptcies, and all dividend omissions

Two interesting questions 1. How often do dividend omissions precede broader defaults? 2. What is the average time from omission to broader default?

84

Dividend Omissions and Broader Defaults

70% 60% 50% 40% 30% 20% 10% 0% Ba B Caa Senior Unsecured Rating at Time of Dividend Omission

Did not default Defaulted later

Median (average) time to further bond default or bankruptcy was about 12 (18) months
85

Preferred Stock Default Rates over a 5-year Horizon are Comparable to Bond Default Rates
60% 50% 40% 30% 20% 10% 0% Aa A Baa Ba B Caa-C
86

Bonds Preferred Stock Preferred Stocks including Div Omission

Issuer senior unsecured rating

Recovery Rates Prices at Default for Preferred Stocks are a Function of the Type of Default Event

Recovery Rates Default Event All Dividend Omissions Omissions Leading to Default Broader Default Mean 36% 34% 15% Median 32% 32% 9% # of Obs 101 41 106

When omissions lead to defaults, recoveries later become roughly the same as those in the broader default category Market does not anticipate which omissions lead to further default
87

Recovery Rates for Defaulted Preferred Stocks in the Utilities Industries Somewhat Higher, But Anecdotally Ultimate Recoveries are Much Higher

50% 40% 30% 20% 10% 0% Dividend omission

Utility Non Utility

Dividend omission further default

Broader default

88

Trust Preferred Stock Defaults

During the same period (1980-2005), 32 issuers defaulted on trust preferred stocks & hybrids Defaults mostly non-financial, non-utility issuers Most of hybrid were either broad defaults or cross defaults Median recovery rate was 11%, slightly lower than that for straight preferred stocks Though most hybrids were backed by jr. sub. notes, their recoveries were more in line with traditional preferred stock than sub debt
89

Summary and Conclusions

Most preferred stocks are investment-grade rated From 1980 to 2005, 268 issuers defaulted About 45% of the defaults were dividend omissions, of which less than 30% were resolved favorably and over 40% preceded broader defaults Preferred stock default rates across various rating categories are fairly comparable to bond default rates thus there is likely no selection bias

90

Summary and Conclusions

Dividend omissions defaults have fairly low price at default, reflecting market expectations that these defaults are unlikely and may precede broader defaults Recovery rates on dividend omissions do not a priori distinguish between dividend resumption and further default Post-default prices on defaulted preferred stocks are higher for utilities, but not by a lot Most of the 32 trust preferred defaults were part of broader defaults, and thus resulted in recovery rates comparable to that for ordinary preferred than to those for subordinate bonds
91

Q&A

92

 Copyright 2006, Moodys Investors Service, Inc. and/or its licensors including Moodys Assurance Company, Inc. (together, MOODYS). All rights reserved. ALL INFORMATION CONTAINED HEREIN IS PROTECTED BY COPYRIGHT LAW AND NONE OF SUCH INFORMATION MAY BE COPIED OR OTHERWISE REPRODUCED, REPACKAGED, FURTHER TRANSMITTED, TRANSFERRED, DISSEMINATED, REDISTRIBUTED OR RESOLD, OR STORED FOR SUBSEQUENT USE FOR ANY SUCH PURPOSE, IN WHOLE OR IN PART, IN ANY FORM OR MANNER OR BY ANY MEANS WHATSOEVER, BY ANY PERSON WITHOUT MOODYS PRIOR WRITTEN CONSENT. All information contained herein is obtained by MOODYS from sources believed by it to be accurate and reliable. Because of the possibility of human or mechanical error as well as other factors, however, such information is provided as is without warranty of any kind and MOODYS, in particular, makes no representation or warranty, express or implied, as to the accuracy, timeliness, completeness, merchantability or fitness for any particular purpose of any such information. Under no circumstances shall MOODYS have any liability to any person or entity for (a) any loss or damage in whole or in part caused by, resulting from, or relating to, any error (negligent or otherwise) or other circumstance or contingency within or outside the control of MOODYS or any of its directors, officers, employees or agents in connection with the procurement, collection, compilation, analysis, interpretation, communication, publication or delivery of any such information, or (b) any direct, indirect, special, consequential, compensatory or incidental damages whatsoever (including without limitation, lost profits), even if MOODYS is advised in advance of the possibility of such damages, resulting from the use of or inability to use, any such information. The credit ratings and financial reporting analysis observations, if any, constituting part of the information contained herein are, and must be construed solely as, statements of opinion and not statements of fact or recommendations to purchase, sell or hold any securities. NO WARRANTY, EXPRESS OR IMPLIED, AS TO THE ACCURACY, TIMELINESS, COMPLETENESS, MERCHANTABILITY OR FITNESS FOR ANY PARTICULAR PURPOSE OF ANY SUCH RATING OR OTHER OPINION OR INFORMATION IS GIVEN OR MADE BY MOODYS IN ANY FORM OR MANNER WHATSOEVER. Each rating or other opinion must be weighed solely as one factor in any investment decision made by or on behalf of any user of the information contained herein, and each such user must accordingly make its own study and evaluation of each security and of each issuer and guarantor of, and each provider of credit support for, each security that it may consider purchasing, holding or selling. MOODYS hereby discloses that most issuers of debt securities (including corporate and municipal bonds, debentures, notes and commercial paper) and preferred stock rated by MOODYS have, prior to assignment of any rating, agreed to pay to MOODYS for appraisal and rating services rendered by it fees ranging from $1,500 to $2,300,000. Moodys Corporation (MCO) and its wholly-owned credit rating agency subsidiary, Moodys Investors Service (MIS), also maintain policies and procedures to address the independence of MISs ratings and rating processes. Information regarding certain affiliations that may exist between directors of MCO and rated entities, and between entities who hold ratings from MIS and have also publicly reported to the SEC an ownership interest in MCO of more than 5%, is posted annually on Moodys website at www.moodys.com under the heading Shareholder Relations Corporate Governance Director and Shareholder Affiliation Policy.

93