Securitization: a Post Mortem

Erik Heitfield Federal Reserve Board erik.heitfield@frb.gov

The views expressed here are my own and do not reflect the opinions of the Federal Reserve Board of Governors or its staff.

Todays Talk
Overview of securitization Looking back What went wrong?
Excessive complexity Leveraged exposure to systematic risk Model error

Looking forward
Alternatives to securitization Securitization going forward
Heitfield -- Federal Reserve Board 2

Overview

What is credit securitization?

Securitization is the pooling and tranching of credit risk Assets of a securitization are a pool of fixed income securities with embedded credit risk such as
Corporate bonds & loans Mortgages Credit card receivables Other structured products

Assets

Liabilities

Super Senior

Liabilities are structured in tranches ordered in terms of payment priority

Senior tranches bear least risk but carry lowest interest rate Mezzanine tranches bear more risk in return for higher rate Lowest tranche (equity) bears most risk and is often not traded

Senior Mezzanine Equity

Heitfield -- Federal Reserve Board

A taxonomy of securitization products

Assets Whole Loans or Unstructured Bonds Agency MBS CDX Indexes Non-agency MBS Corporate Bond CDO Structured Securities ABX.HE Indexes ABS CDO CDO-squared

Liabilities

Pass-through

Structured

Heitfield -- Federal Reserve Board

Goals of securitization
Separate debt funding from risk bearing
Senior tranche investors put up most of the capital to fund assets Junior tranche investors bear most of the credit risk

Risk diversification
A single structured credit deal may be backed by hundreds or thousands of loans Cash flows for large asset pools may be easier to predict

Tailor debt risk/return characteristics to market demand

Risk/return profile of structured securities is different from that of underlying collateral

Regulatory arbitrage
Financial institutions can move assets off balance sheet while continuing to be exposed to much of those assets credit risk
Heitfield -- Federal Reserve Board 6

Rapid growth of new MBS

Annual Issuance of Non-agency Residential Mortgage Backed Securities
$ billions (nominal) 1,400 Other Prime Alt-A Subprime

1,200

1,000

800

600

400

200

0 2001 2002 2003 2004 2005 2006 2007 2008*

* Jan-Sep, Annualized Source: Inside Mortgage Finance

Heitfield -- Federal Reserve Board

Increasingly held by CDOs

Composition of CDO Collateral
High-Grade Cash-Flow/Hybrid CDOs
High-Grade Cash-Flow/Hybrid CDOs
100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 2001 2002 2003 2004 CDO Vintage
* 2007 vintage includes deals completed through September Source: Standard and Poors

Mezzanine Cash-Flow/Hybrid CDOs

Mezzanine Cash-flow/hybrid CDOs
100% 90% 80% 70%
Other 60% CDOs NIMS 50% Closed-end 2nds Alt-A 40% Subprime
Other CDOs NIMS Closed-end 2nds Alt-A Subprime

30% 20% 10% 0%

2005

2006

2007*

2001

2002

2003

2004 CDO Vintage

2005

2006

2007*

* 2007 vintage includes deals completed through September Source: Standard and Poors

Heitfield -- Federal Reserve Board

MBS valuations plunge

AAA ABX.HE Indexes
110

100

90

80

70

2006:H1 2006:H2 2007:H1 2007:H2

60

50

40 Jan-06

Jul-06

Jan-07

Jul-07

Jan-08

Jul-08

Source: MarKit

Heitfield -- Federal Reserve Board

ABS CDO credit ratings collapse

Share of CDO notes downgraded to Caa or lower, by initial grade and vintage Percent 90
80 70 60 50 40 30 20 10 0 Aaa
Source: Moody's

2007 Vintage

2006 Vintage

Aa Initial Rating

Baa

Heitfield -- Federal Reserve Board

10

and so does CDO issuance

Global CDO issuance
$ Billions 200 180 160 140 120 100 80 60 40 20 0 05-Q1 05-Q2 05-Q3 05-Q4 06-Q1 06-Q2 06-Q3 06-Q4 07-Q1 07-Q2 07-Q3 07-Q4 08-Q1 08-Q2

Unstructured Collateral (corporate bonds, loans, etc.)

Structured Collateral (MBS, other ABS and CDOs)

Source: Securities Industry and Financial Markets Association

Heitfield -- Federal Reserve Board

11

What went wrong?

Excessive complexity Leveraged exposure to systematic risk Sensitivity to modeling error

Heitfield -- Federal Reserve Board

12

Complexity

Tasman CDO
$300 million mezzanine-hybrid CDO2 Deal date: January 11, 2007 Lead Underwriter: UBS Capital Structure: 7 debt classes maturing in March 2047 Assets: 64 CDO notes of various types

Heitfield -- Federal Reserve Board

14

Tasman CDO Liabilities

Class Name A1S Senior Secured A1J Senior Secured A2 Senior Secured A3 Secured Deferrable B Mezzanine Secured Deferrable C Mezzanine Deferrable U.S Subordinated
Maturity: 03/21/2047

Issue Amount ($MM) $164 $30 $58 $20 $12 $4 $12

Initial Rating Aaa Aaa Aa2 A2

A1J Senior Secured 10%

A1S Senior Secured 55%

Baa2 Ba1 NR

A2 Senior Secured 19% A3 Secured 6.66% B & C Mezzanine Secured 5.33% U.S Subordinated 4% 15

Heitfield -- Federal Reserve Board

Tasman CDO Assets

TASM (CDO2) TASM (CDO2)

CAMBER-8 CAMBER-8
(CDO2) (CDO2) 0.72% 0.72%

AQUARIUS-4 AQUARIUS-4
(CDO2) (CDO2) 0.58% 0.58%

62 Other 62 Other CDO Notes CDO Notes 98.7% 98.7%

PINE-5 PINE-5
(CDO2) (CDO2) 0.004% 0.004%

MAY-5 MAY-5
(CDO2) (CDO2) 0.94% 0.94%

Merrill Lynch Loan Merrill Lynch Loan

(Home Equity) (Home Equity) 0.005% 0.005%

Other Other Debt Notes Debt Notes

AQUARIUS-5 AQUARIUS-5
(CDO2) (CDO2) 0.19% 0.19%

Other Other Debt Notes Debt Notes 16

Heitfield -- Federal Reserve Board

Tasman CDO Assets

CDO Direct CLO Other ABS Corporate CMBS Prime/Midprime Indirect Alt-A Subprime Second 0% 20% 40% 60% 80% 100% HELOC

Note: Indirect exposure tabulated at three-level depth.

Heitfield -- Federal Reserve Board

17

Tasman CDO Performance

Tranche Credit Ratings (21 Notch Scale)
Aaa

20 A1S 15 A1J A2 A3 B C

Baa

10
Caa

5 0 1/11/07 1/30/08 3/07/08 6/16/08 7/09/08

Entered accelerated repayment on March 17, 2008

Heitfield -- Federal Reserve Board 18

Leveraged exposure to systematic risk

Pooling assets does not reduce systematic risk

Performance of collateral assets depends on two types of risk factors
Idiosyncratic factors unique to each asset (e.g., quality of a firms management, homeowners individual financial condition) Systematic factors shared by all assets (e.g., macro environment, aggregate home price appreciation)

Pooling assets limits importance of idiosyncratic risk

Law of Large Numbers implies that loss rate for a pool of securities is less volatile than the loss rate for an individual security

But pooling assets does not diminish systematic risk

Systematic risk factors induce correlations in losses across securities

Loss rate for a large pool of securities has less dispersion overall, but systematic factors play a bigger role
Heitfield -- Federal Reserve Board 20

Large asset pools may not be safer under systematic stress conditions
Loss Exceedance Probabilities (%) Num. of Bonds 20 200 20 200 Loss > 5% 18.6 16.3 67.2 81.2 Loss > 10% 6.3 3.1 39.4 37.5 Loss > 15% 2.0 0.5 16.5 11.6

Unconditional

Systematic Stress*

* Assumes 99th percentile draw of systematic risk factor

Heitfield -- Federal Reserve Board

21

Senior tranches are sensitive to macro shocks

Senior tranches are always safer than junior tranches of the same CDO But credit performance of senior tranches is more sensitive to systematic shocks
Expected Losses (%) for Simulated CDO Tranches Tranche Collateral Pool 3 - 6% 6 - 9% 9 - 12% 12 - 100% Unconditional Expected Loss 3.5 10.4 3.6 1.2 0.0 Stress Condition Expected Loss* 11.7 72.0 39.8 20.4 0.5 Ratio 3.4 6.9 11.5 16.6 35.5

* Stress condition assumes 99th percentile draw of systematic factor.

Ratio of Stess Expected Loss to Unconditional Expected Loss Simulated CDO of 200 Mezzanine Bonds
40 35 30 25 20 15 10 5 0 Collateral Pool 3-6% Tranche 6-9% Tranche 9-12% Tranche 12-100% Tranche

Heitfield -- FederalEL assumes 99th percentile draw of a systematic risk factor. Reserve Board Stress

22

Model risk

Senior tranche risk metrics are more sensitive to model errors

Simulated CDO tranche probability of default (PD), expected loss (EL), and conditional expected loss (EL0.99) as a function of collateral default probability parameter () and asset correlation parameter ().

Heitfield -- Federal Reserve Board

24

so senior tranches are harder to evaluate given limited data

Simulated distribution of CDO tranche probability of default (PD), expected loss (EL), and conditional expected loss (EL0.99) given historical data on the performance of 20 cohorts of 400 collateral bonds.

Heitfield -- Federal Reserve Board

25

Conclusions
To a large degree, the potential for dramatic falls in ratings and valuations was baked in to the securitization process
Complexity of exotic resecuritization deals made it difficult to evaluate risk exposures Pooling assets and structuring liabilities ensured that senior tranches would perform well under most circumstances, but would all fall together Credit rating and valuation models depended on limited data could not accurately assess the likelihood of future losses
Heitfield -- Federal Reserve Board 26

Looking Forward

Regulators and market participants are working to address identified problems

Originators, underwriters, and sponsors created increasingly complex structured products, but did not always supply investors with sufficient information on the assets backing them Rating agencies underestimated the risk of RMBS and complex resecuritization products Financial regulators did not adequately distinguish between structured and unstructured credit exposures Investors, attracted to higher yields, relied on credit ratings, ignored exposure to systematic risk
Heitfield -- Federal Reserve Board 28

Alternatives to structured MBS are expanding

Figure extracted from Best Practices for Residential Covered Bonds, US Treasury Dep.

Heitfield -- Federal Reserve Board

29

Covered bonds
US regulators are promoting use of covered bonds to finance mortgage lending (July 2008)
FDIC clarified receivership treatment of covered bonds Treasury articulated best practice standards for covered bonds

Covered bonds offer an alternative funding source for banks

More flexible than deposits Cheaper than unsecured bank-issued bonds

Unlike securitization, covered bonds do not transfer credit risk

If collateral assets perform poorly, bank equity-holders bear first losses, not covered bond investors Collateral performance only affects bond returns if the issuing bank defaults

Effects on bank balance sheets

No direct effect on assets pledged collateral stays with the bank May increase risk to FDIC, uninsured depositors, and unsecured bond-holders May make it easier for banks to manage liquidity
Heitfield -- Federal Reserve Board 30

Will securitization come back?

Greater emphasis on transparency and standardization
Example: standardization of credit default swap terms and central clearing of CDS contracts

Rating agencies and other gatekeepers have become much more conservative Some structured finance products including resecuritizations such as ABS CDOs and CDO2 may no longer be viable
Heitfield -- Federal Reserve Board 31