Cdos

Collateralised Debt Obligations Copyright 2004 Deutsche Bank@

Collateralised Debt Obligations

Balance Sheet and Arbitrage CDO’s

Mike Pawley [email protected]

Global Markets Training 1



This material is divided into two sections : balance sheet CDO’s and arbitrage CDO’s (page 25). This first session explains a specific sector of the CDO market : balance sheet CLO’s (or Collateralised Loan Obligations, CLO’s).

What will you get out of this material ?

Well, at the end of this session you should :

Understand the rationale for issuing balance sheet CDO’s

Be aware of the credit enhancements for balance sheet CDO’s

What you already need to know:

You should already understand :

The basic mechanics of ABS.




We start with a very general overview of the CDO market.

The term Collateralised Debt Obligation (CDO) is a general name for investment vehicles backed by loans, bonds, structured products, or various mixtures of all three.

Securities backed by bank loans are called Collateralised Loan Obligations (CLO’s). Securities backed by bonds are known as Collateralised Bond Obligations (CBO’s). More recently, CDO’s have evolved investment vehicles that are backed by structured products (surprisingly enough, called Structured Product CDO’s or ABS CDO’s).

The terminology tends to overlap to some extent. For example, some deals that are called CBO’s may contain loans as well as bonds in the asset pool. Try not to get too carried away with strict definitions – the product terms tend to blur at the boundaries somewhat.

The figure below shows a useful categorisation of the products and also a time tag for each product to give you some idea as to how the product has developed over the last fifteen years or so. The list is not meant to be exhaustive – new innovations are occurring at a rapid pace.

CDO Product Range


CDO’s

CBO’s CLO’s

Cash Arbitrage CBO’s1988 on

Market Value Arbitrage CDO’s

Balance Sheet CLO’s1995 on

Synthetic Balance Sheet CLO’s1997 on

Synthetic Arbitrage CDO’s1998 on

Single TrancheCDO’s2001 on

Index CDO’s2003 on

CMCDO’s2004 on

CDO’s

CBO’s CLO’s








CMCDO’s2004 on


In a conventional CDO, bonds/loans/assets are sold to a special purpose vehicle (SPV). The SPV issues notes to investors that are backed by the collateral. The interest on the collateral is used to pay the interest on the investors’ notes, whilst the principal on the collateral eventually pays off the investors’ principal.

The notes are typically in tranches with ratings from AAA to B and an unrated ‘equity’ slice that is often kept by the originator (see the figure below).

Conventional CDO

‘Tranching’ refers to the issuance of securities that are successively subordinated to one another. The credit enhancement to achieve AAA ratings comes from this note tranching and subordination. In other words, the most junior (equity) tranche takes the first losses in the collateral pool. If the most junior tranche is wiped out by losses, then the next most junior tranche starts to take losses, all the way up the credit spectrum.

The securities also tend to be ‘sequential pay’. In other words, principal and interest is paid to the highest rated notes first (which is also a form of credit enhancement).

Any interest flows left over after the noteholders have received their cut is returned to the originator (typically after any defaults). This allows the originator to increase return on equity (ROE) because a spread is being earned despite the assets being off-balance sheet. It is also a way of releasing capital and achieving regulatory capital relief (see the ABS session for more detail).

Some of the terminology in this area can be confusing. The figure below attempts to explain the differences between attachment points, tranche thickness and credit enhancement (subordination).



CDO Terminology

Conventional CLO

Here’s an example to show a conventional balance sheet CLO.

CORE 1999-1 LTD (Deutsche Bank AG)

CORE provides the opportunity to invest in the Mittelstand, small and medium sized German corporates. Very few of these companies have debt outstanding in the capital markets and few are listed – CORE provides pooled exposure to over 3,700 companies, effectively allowing investors to buy the German economy. The loans are all fixed rate.

Group Treasurer Detlef Bindert commented at the time, “It is very significant for us, because it allows us to roll assets off our balance sheet to make room for new transactions.”

Deutsche were aiming to make a 25% return on regulatory capital by 2001 (its return on capital was below 10% in 1999). None of the loans has ever been in arrears or required bad debt provisioning.

Credit enhancement comes from around 180bp of excess spread, and a reserve fund that builds from 1.3% to 2.3%, fed by excess spread. The €32m of funding for the excess reserve is now Deutsche’s only capital commitment to the portfolio.

“We want to improve our return on regulatory and economic capital through securitisation, and we have already begun homogenising our lending practices to make securitisation easier,” said Jurgen Bilsten, board member of the CORE (Corporate and Real Estate) division.

The deal was about 1.8 times oversubscribed.


Equity 4%

BBB 5%

AA 14%

AAA 77%

0%

4%

100%

9%

23%

0%

4%

9%

23%

Attachment Points

Credit enhancement or Subordination

Tranche Thickness

Equity 4%

BBB 5%

AA 14%

AAA 77%

0%

4%

100%

9%

23%

0%

4%

9%

23%

Attachment Points

Credit enhancement or Subordination

Tranche Thickness


CORE TRANCHES

Class Size Rating Exp. Maturity

Coupon

A1A €365.8m P-1/A-1+ Dec 1999 3m Euribor + 6bp

A1B $194m P-1/A-1+ Dec 1999 3m Euribor + 3.5bp

A2 $274.4m Aaa/AAA Jun 2000 3m Euribor + 8bp

A3A €1.0341bn Aaa/AAA Jun 2004 3m Euribor + 21bp

A3B $250m Aaa/AAA Jun 2004 3m Euribor + 19.5bp

A4 €231.1m Aaa/AAA Dec 2005 Fixed 3.975%

M €44.7m Aa2/AA Dec 2005 3m Euribor + 30bp

B1 €49.6m A2/A Dec 2005 3m Euribor + 45p

B2A €20.6m Baa2/BBB Dec 2005 3m Euribor + 100bp

B2B €29.0m Baa2/BBB Dec 2005 Fixed 4.185%

B3A €7.9m Ba2/BB Dec 2005 3m Euribor + 300bp

B3B €47.0m Ba2/BB Dec 2005 Fixed 6.705%

CORE 1999-1 included three US dollar denominated tranches totaling $718.4m to attract US investors, but the balance of demand made CORE predominantly a European trade. Deutsche’s name attracts a premium in Europe, and investors are familiar with the risk of the German economy.

“One of our objectives each time we bring a deal is to grow the investor base,” said a spokeswoman for Deutsche’s European securitisation group.

As currency and interest plays have vanished with the single currency, many investors are looking to the ABS market as a safe source of yield. In particular, asset backed securities do not have the event risk associated with sovereigns and corporates. In Europe, at the worst point of the credit crisis of 1998, spreads on AAA securitisations only widened by a couple of basis points and BB tranches only went out by 40 or 50bp.

While many CLO issuers have striven to fashion bullet bonds, Deutsche’s policy is to pass amortisation through to investors and provide multiple tranches to fit interest rate and maturity needs for particular categories of investor.

The deal included some fixed rate tranches (unusual for a CLO) for new investors who wanted that format.

The A1 tranches, with short lives, went to money market accounts, particularly in UK, France, Germany, and Belgium. Deutsche designed the A2 tranche in dollars in response to reverse enquiry from US money managers.

The majority of the deal came in the A3A and A3B tranches with expected maturities of around five years. Deutsche Bank had over fifty investors in these tranches, spread over the world and across insurance companies, pension funds, asset managers, hedge funds and banks.

The CORE deal above was quickly followed by CORE 1999-2.



CORE 1999-2 Details

The CORE 99-2 transactions securitised 2,958 corporate loans with maximum remaining maturity of 98 months, to over 2,281 borrowers. Within the collateral, no onesingle borrower accounted for more than 1.5%of the portfolio by principle balance.

The figures below show the distribution of sales for CORE 1999-2.

Investor Geographic Distibution

Investor Type Breakdown


US25%

Belgium3%

France13%

Ireland5%

Other3%

UK15%

Germany35%

Luxem.1%



Funds62.7%

Insurance

30.9%

Corporate6.4%


Credit Derivatives and Securitisation

Credit derivatives are now common in CDO’s. CDO’s that have pools of credit default swaps as the collateral pool are known as ‘synthetics’. It is instructive to see how synthetics developed over time :

Early Synthetic CLO’s (late 1990’s)

In early examples of synthetic CLO’s the assets were not sold to the SPV, but instead remained on the originating bank’s balance sheet. However, the credit risk of the assets was transferred to the SPV using a portfolio credit default swap. The SPV issues tranches of notes to investors (see the figure below). The proceeds received from the investors are used by the SPV to buy high grade collateral (typically AAA government bonds or AAA pfandbriefe).

Investors’ interest is paid for out of the fee from the portfolio credit default swap and the interest on the collateral pool. Any excess spread is returned to the originator. At maturity, the collateral pool is used to repay principal to the note investors.

One of the key benefits of these early synthetic transactions is that they avoid many of the problems associated with transferring assets to the SPV (client permission, tax problems etc).

Early Synthetic CLO

The table below names some of the more well known CDO deals and their structures.

Early Basic Synthetic Structures

1. SPV, Fully Funded, Conventional; CORE 1999-1,1999-2 2. SPV, Fully funded, Synthetic; Geldilux 1999-1, Europa Two (CMBS)3. SPV, Partially Funded, Synthetic, Super-Senior Tranche; BarCLO 1999 4. No SPV, Partially Funded, Synthetic, Super-Senior Tranche, CLN; CAST 1999-1, 2000-1, GLOBE 2000-1

The Geldilux example below demonstrates many of the key principles.



Geldilux

In 1999 HypoVereinsbank AG (HVB) announced a structured CLO transaction.

Geldilux Key Terms

Issuer: GELDILUX 99-1 Limited (SPV)

Principal: €2.1 bn (including USD500m)

Structure: Senior/subordinated structure in 6 tranches

Maturity: 3 years

The issuer is Geldilux 99 - 1 limited, a special purpose company/vehicle (SPV) owned by a charitable trust. Geldilux is a bankruptcy-remote special purpose company whose only activity will be to issue and service the notes.

The underlying loan portfolio is held by HypoVereinsbank, Luxembourg (HVL).

HVL is a wholly owned subsidiary of HVB. HVB was formed in November 1998 from the merger of Bayerische Vereinsbank AG and Bayerische Hypotheken-und Wechselbank AG.

The figure below shows the key elements of the structure :

Geldilux Structure

This CLO is unusual because HVL does not physically sell the loans to Geldilux. It only transfers the economic risk of the reference pool. Geldilux issues a guarantee for the economic performance of the reference portfolio to HVL, which establishes the transfer of risk. HVL can draw under this guarantee if loan write-offs have occurred in the reference loan portfolio.



The rationale for doing a synthetic CLO in this case was quite simple. HVB did not want to alienate corporate clients with a conventional CLO where HVB would have to ask permission to sell the loans. The message from HVB is that they stay with their core business of lending and they do not intend to stop servicing and analyzing their clients.

The guarantee issued by Geldilux has an effect similar to a credit default swap. To pay for the guarantee, HVL pays a fee, like in a credit default swap. Under the guarantee HVL may draw once an event of default has occurred. Unlike a regular credit default swap, the guarantee may be used again if an additional event occurs. Thus the guarantee resembles a pool of credit default swaps on each loan in the reference pool.

Credit Default Swap

Collateral Pool

In a conventional CLO, the SPV uses the proceeds of the sale of notes to buy the loans. In this early synthetic case, Geldilux uses the funds to buy collateral that backs up its guarantee to HVL.

The proceeds from the sale of the notes will be used to:

Purchase floating rate Pfandbriefe issued by HVB. The Pfandbriefe notes are rated AAA.

Invest in Euribor deposits at HVL or HVB

The proceeds from the sale of the A-1 and A-2 class are used to purchase the Pfandbriefe. Pfandbriefe make good collateral because it is AAA rated. Also, the Pfandbrief is a floating rate issue thus minimizing interest rate risk. Geldilux will always be able to sell it near par unless HVB’s credit spread moves out.

HVB Risk Management Products (wholly owned subsidiary of HVB) has sold to the issuer a put option that the issuer can exercise in case the market price of the Pfandbrief collateral does indeed fall below par. If the value of the Pfandbriefe falls below par at any reset date or at redemption, the put option will pay Geldilux the difference between the value of the Pfandbriefe and par. If HVB is downgraded below a rating of A, the put options are automatically exercised.

Proceeds from sale of B, C, D and E classes are placed in Euribor deposits.



The guarantee will be used to absorb losses on the loans. The ranking of the guarantee is based on class. The E class absorbs any losses first up to the C class.

Geldilux Tranches

Currency Principal Class Maturity Rating

EUR 1,571,073,000 A-1 Feb 02 AAA

USD €equiv 476,190,476 A-2 Feb 02 AAA

EUR 55,319,000 B Feb 02 A

EUR 21,276,000 C Feb 02 BBB

EUR 31,914,000 D Feb 02 BB

EUR 19,151,000 E Feb 02 NR

Total 2,174,923,476

The A-2 class pays note holders in USD. Geldilux will enter into a cross currency swap for the principal of this class. The counterparty is Goldman Sachs Mitsui Marine Derivative Products LP.

The tranching gives the following credit enhancement :

Credit Enhancement

A 6%B 3.5%C 2.5%D 1.0%E -

The A1 and A2 notes (94% in total, all AAA) therefore have 6% protection.

Geldilux Subordination

Class Rating Comments

E NR Deposits used to pay any loan losses

D BB Deposits used to pay losses only after Class E has been retired

C BBB Deposits used to pay losses only after Classes E and D have been retired

B A Deposits used to pay losses only after Classes E, D and C have been retired

If all of the deposits above have been used to cover loan defaults, Pfandbriefe will be sold to cover any further defaults.



Geldilux will receive a servicing fee from HVB for this loan guarantee.

Principal is scheduled to be repaid at maturity in February 2002. Unlike structures from other issuers (i.e. Deutsche Bank AG’s CORE) the investors do not carry any prepayment risk, as the loans are not sold to the SPV.

However, as part of the credit enhancement, notes are redeemed immediately if:

HVL or HVB is insolvent

HVB defaults on Pfandbriefe

Interest on the notes will be funded by interest from the collateral (deposit and Pfandbrief issue).

The fee HVL pays to Geldilux makes up for the difference between the interest received from the collateral pool and the interest distributed to the noteholders. Allianz Risk Transfer NV guarantees the payment of this fee up to €15m.

Interest is paid quarterly in arrears. Interest is paid at 3-month Euribor + spread on each class.

Geldilux Spreads at Issue

Class Rating Libor spread

A-Class AAA 20 bps

B-Class A 40 bps

C-Class BBB 100 bps

D-Class BB 250 bps

E-Class N.R Not disclosed

Linked and Delinked Structures

The terms “linked” or “delinked” refer to the credit rating of the CLO securities issued in relation to the credit rating of the bank selling the loans. This is one of the major structural distinctions among CLOs, which is also reflected in the pricing and relative value considerations.

Geldilux is interesting in that the A-classes are delinked and the other classes are linked to the credit rating of HBV.

The class B, C and D notes are linked to the HVB rating. If HVB’s unsecured rating is downgraded below the then-outstanding rating of the class B, C and D notes these notes will be downgraded accordingly.

At maturity, principal on the notes is funded through the redemption of the Pfandbrief and the deposit.

Early redemption of the notes occurs if:

HBV fails to pay interest on the Pfandbrief;

HBV or HLV enter bankruptcy procedures.

In that case, the guarantee of Geldilux towards HVL extinguishes



In case HVB defaults under its obligation under the put option and the trustee is not able to sell the Pfandbriefe at par, A class note holders receive delivery of the Pfandbriefe.

If the rating of HVB falls below A HVB Risk Management must take steps to eliminate the exposure to HVB. This might be done through substitution of HVB Risk Management

These measures, combined with the put option on the pfandbriefe, make the collateral support for the “A” classes very strong and delinked from the rating of the selling bank.

HVL itself has no rating but has obtained general support from HBV. So, the ratings of the classes B - D are related to the rating of HBV, as any loss in the deposit will be fully attached to these classes.

Though the loans stay on HVL’s (HVB’s) balance sheet they no longer require any capital. The relief of core capital amounts to more than €100m, according to the board of HVB.

Geldilux Structure

HVL owns a large number of loans made to HVB’s corporate and private clients. According to certain limiting criteria loans are put together into a defined pool of loans (the reference pool).

The portfolio guaranteed by Geldilux contains 1,355 loans.


Geldilux 99

Pfandbriefe,DepositsDeposits

Interest

Put Option onPfandbriefe

Currency SwapA2 Notes

EUR

HVL

Guarantee

Fee

Notes

USD

Interest,Fee


Reference Pool

Criteria

Size Of Pool EUR 2.2 bn

Max maturity 364 days

Internally Rated Yes

Internal Rating <8

The internal loan rating system rates loans from 1-10. One is the best rating and 10 is the worst rating.

Most of the loans are to small and midsize corporates or to private clients.

Clients: Midsize corporates and private clients

Single obligor limitation: Single borrower must not exceed 1% of portfolio.

Maturity limitation: Maturity no longer than 364 days

Actual average maturity: 105 days

Rating: Loans are unrated but are only handed out to clients with long standing relationships and favorable track record.

Single biggest loan: EUR 20m.

Loans to Bavarian borrowers:

56.9% of portfolio

Loans to real estate sector:

29% of portfolio. Loans to this sector could represent up to 45% of the reference pool. If certain loss limits are reached, substitutions to the pool can’t include real estate loans until their exposure has been reduced to 20%.

Loans to private clients: 26% of portfolio

FitchIBCA could not apply their loan rating methodology to the portfolio because none of the loans had external ratings to map to the internal ratings.

Loans were not externally rated because of their size and type of customer.

S&P used historical losses over a 10-year period (1988 –1998) on HVB’s loan portfolio to rate the transaction.

S&P considered:

Diversity of the loan portfolio

Number of loans in the portfolio

Quality of HVB’s internal rating system.

S&P consider the portfolio to be diversified except for some concentration to Bavaria and in some real estate sectors. This has been compensated for through limiting criteria regarding replacement of loans.



Synthetic CLO’s With Super-Senior Tranches

The next development in the evolution of CDO’s came with the use of ‘super-senior’ tranches, as in the figure below. This type of synthetic structure permits the sponsoring institution to obtain credit protection on a much larger reference portfolio than would be practicable in the conventional CLO market.

There is credit leverage in a synthetic CLO because the reference portfolio is much larger than the note issue. The synthetic structure thus allows the issuer to obtain credit protection on a much larger reference portfolio than would be practicable in the conventional CLO format.

Conventional CLO’s do not offer this element of credit leverage, and have a one-to-one relationship between the collateral pool and the note issue.

Synthetic CLO with Super Senior Tranche

In effect, only a small part of the asset pool is securitised through the sale of notes. The reference portfolio, in other words, is typically much larger than the note issue.

The credit risk of the larger part of the portfolio (the super-senior part) is typically transferred via a credit default swap. This also makes it easier to create notes that have (soft) bullet repayment. It appears that many investors prefer soft bullets, so it is likely that this type of structure may become more prevalent in the market.

The synthetic deal with a super senior tranche tends to be cheaper than a conventional structure and hence gives greater returns to equity, as can be seen in the figure below.



You can see from the spreadsheet below that the return to the equity holders in a conventional CDO is 12.22%, whereas for the synthetic it is 18.81%. This is largely because of the cheapness of the super-senior default swap.

Go to spreadsheet ‘CDOs – Synthetic’ to see the calculations.


AAA(75-80% ofportfolio)

SuperSenior

Equity

AA

Equity

AA

AAA

Libor+35 to 50

Libor+100 to 125

10 to 20 bps

Libor+40 to 65

Libor+100 to 125

Super SeniorConventional

AAA(75-80% ofportfolio)

SuperSenior

Equity

AA

Equity

AA

AAA

Libor+35 to 50

Libor+100 to 125

10 to 20 bps

Libor+40 to 65

Libor+100 to 125

Super SeniorConventional

Portfolio 1,000Return 2.30%

Conventional SyntheticTranche Rating Notional Percentage Spread Notional Percentage SpreadSS - 750 75.0% 0.15%A AAA 770 77.0% 0.45% 110 11.0% 0.45%B AA 140 14.0% 1.00% 50 5.0% 1.00%C BBB 50 5.0% 3.00% 50 5.0% 3.00%

Total 960 0.64% 960 0.362%

Portfolio Income 23,000,000 23,000,000 Portfolio Funding Cost 6,110,400 3,475,200 Expenses and Losses 12,000,000 12,000,000 Difference 4,889,600 7,524,800

Equity 40Equity Spread 12.22% 18.81%


Example : BarCLO Finance (1999) Ltd, Synthetic CLO

Barclays Capital securitised a loan portfolio using similar techniques to that above. BarCLO is an SPV incorporated in the Caymans, created to issue the notes listed below and to enter into certain transactions with Barclays Bank PLC.

BARCLO NOTES

Note Class Amount Type Maturity Rating

A $65m FRN 2004 AAA

B $60m FRN 2004 A

C $35m FRN 2004 BBB

D $20m FRN 2004 BB

E $20m FRN 2004 NR

The loans are not transferred to the SPV. Instead, in this early synthetic CLO, the credit risk is transferred by Barclays through two credit default swaps.

The issuer (BarCLO) has entered into a credit default swap with Barclays Bank PLC under which the issuer agrees to make payments to Barclays (up to a maximum of $200m) for losses on a pool of reference credits with a notional amount of $2bn. All credits in the pool will have investment grade ratings by public rating agencies.

The noteholders have security in the form of the repurchase agreement (repo), repo collateral, allowable investments in government securities and the credit swap with Barclays.

The ratings reflect the investment grade quality of the reference loans, the structure of the transaction (including credit enhancement levels), and Barclays’ strength as repo and credit swap counterparty.

In addition, there is a credit default swap with Bank of America that covers any losses over $200m (ie the next $1.8bn of the portfolio) for the super-senior tranche. The default risk of the super-senior class, rather than being retained, has therefore been passed on to the Bank of America. Barclays still needs to hold a 20% capital weighting against a default by Bank of America.

The protection for this (8 basis points) is relatively cheap because losses up to $200m have to occur before Bank of America becomes liable to make default payments to Barclays.



Structure of BarCLO Synthetic CLO

If the obligor under a reference credit fails on a payment of principal or interest in respect of $10m or more or becomes bankrupt a credit event will be deemed to have occurred.

BARCLO NOTE STRUCTURE

A $65m 32.5% AAA

B $60m 30.0% A

C $35m 17.5% BBB

D $20m 10.0% BB

E $20m 10.0% NR

Total $200m 100.0%

The issuer will invest in US treasuries and/or repurchase agreements (US Treasuries) with Barclays. The US treasuries, repo, and all payments due to the issuer under the credit swap will serve as collateral for the noteholders, subject to a first priority lien by Barclays under the credit swap. This structure allows the most senior tranche to be rated AAA, which is above Barclays AA+ rating from Fitch IBCA.

The repo collateral will be marked to market daily and must equal at least 100% of the principal balance of the notes plus accrued but unpaid interest due under the repo. If there is any deficit Barclays will have to post additional securities or cash.



At maturity, Barclays will repurchase the government securities at their market price. The proceeds of this repurchase will be available to repay principal to the noteholders, net of any credit payments due under the credit swap.

It is possible that there may not be market appetite for $2bn of CLO notes. The secondary credit default swap will be cheap because Barclays and the noteholders take the first loss up to $200m. Also, the credit derivative route is cheaper.

Interest on the notes will be paid monthly from proceeds paid by the repo counterparty, income received on permitted investments, and amounts due under the credit swap.

If Barclays fails to make payments under the repo or credit swap an event of default will occur and the repo collateral can be liquidated. Exposure to market value changes of the repo collateral is limited to the five day grace period the repo counterparty has upon default, plus the time it would take to liquidate the securities.

Interest is paid on the full principal balance of the notes, since credit protection payments under the credit swap are not due until maturity of the transaction.

No principal will be distributed to noteholders until the scheduled maturity of the notes (unless there is a ‘termination’ event eg default by swap counterparty). Similarly, Barclays will not receive any credit protection payments until scheduled maturity.

At maturity Barclays will repurchase the repo collateral from the issuer. The cash repurchase price will go first to pay any credit protection payments due under the credit swap, then to repay the notes in order of seniority.

Reference Credit Pool

The default risk being transferred is on a highly diversified pool of investment grade credits. There are 181 obligors in the pool. An obligor with a rating of AA+ or better can compose up to 1.2% of the reference pool; individual obligors rated AA through A- can account for up to 1% of the pool; and any credit rated below A- cannot represent more than 0.5% of the pool. The lowest rating allowed is BBB- and obligors so rated are limited in aggregate to 10% of the total reference portfolio.

Industry concentrations in the pool are limited to 10%, with the top three allowed to exceed 10% but none higher than 15%. Exposure to countries rated below AA is limited to 10%; there can be no exposure to countries rated below investment grade.

The reference credits can reference a whole or part of a credit; Barclays need not own any reference credits and may assign or terminate its right/obligations under any of the reference credits. Credits that are deleted from the pool may be replaced by other reference credits as long as the pool meets the covenants in the transaction documents.

Credit Enhancement and Stress Tests

Credit protection is provided by the subordination of the E notes to the D notes, which are junior to the class C notes which are junior to the class B notes, which are junior to the class A notes.

Based on the maximum reference pool of $2bn the class A notes have credit enhancement of 6.75% (eg $135m/$2bn); class B has 3.75%; class C has 2.00% and class D has 1.00%.

Fitch IBCA stressed the reference credits to default levels ranging from 2.0% to 6.5%. Recoveries were assumed to be 42.5%, well below Fitch IBCA’s experience with bank



loans. The structure passed all stress tests at the rating levels below. Fitch IBCA also ascertained that the rated debt could withstand defaults of several of the largest permitted obligors in the pool without suffering any losses.

The deal significantly enhanced return on weighted risk assets (WRA’s).

Defaulted credits will be put up for three bids between the 20th and 30th day after the credit event has been declared. If at least two bids have not been received the process will be repeated. If two bids are received between the 60th and 90th business day, the higher of these will be considered the value of the loan. If two quotes are not received on either of these dates, a nationally recognised independent accounting firm will value the credit by the 120th day. The credit protection payment due to Barclays will be one minus the value determined by the quotes multiplied by the par amount of the credit. Initially all credits will have this method of default payment.

Example : CAST 1999-1, Synthetic CLO (No SPV), CLN

Here’s a good example of an early synthetic CLO with a super-senior tranche. The structure of the Deutsche Bank CAST deal is shown below.

CAST 1999-1

The tranching is as follows :



The CAST synthetic deal is an extension of Deutsche Bank’s established CORE CLO programme. As mentioned above, synthetic structures permit the sponsoring institution to obtain credit protection on a much larger reference portfolio than would be practicable in the conventional CLO market.

The main element of interest in the CAST structure is that it does not contain an SPV. Instead, the notes are direct obligations of Deutsche Bank, with an embedded credit contingency based on the reference portfolio from Deutsche’s middle market corporate loan portfolio.

As the notes are direct obligations of Deutsche Bank the rating of each note class is linked both to Deutsche Bank’s rating and the exposure to losses in the reference portfolio (see section on ‘Linked’ and ‘Delinked’ in the Geldilux deal above).

The A and B classes (AAA to AA) obtain their high ratings (relative to DB) by being collateralised by AAA rated pfandbriefe issued by Eurohypo AG. In effect, this de-links them from the Deutsche Bank rating.[The pfandbriefe collateralise the remaining outstanding balance of the A and B classes after taking into account any qualifying credit losses in the reference portfolio, not the full initial notional principal on the notes].

Should the issuer default the trustees will sell pfandbriefe at the market price. If the long term unsecured debt rating of the issuer is downgraded below F1 by Fitch IBCA or A-1+ by S&P, the issuer will enter into a put option agreement with a counterparty with at least the equivalent of the above ratings from any two of the rating agencies. The put will require the counterparty to purchase pfandbriefe at par plus accrued interest.

Classes C, D and E do not have the benefit of any collateral. They are therefore linked to the rating of Deutsche Bank and to the performance of the reference pool.

The notes are bullet structures that mature in October 2006 (legal final 2008, two years after the expected maturity to allow for any workout of defaulted loans). There is therefore some extension risk (starting with the most junior notes) although the notes will continue to receive interest in the extension period.



There is the ability to add new loans to the pool (subject to specific rating criteria) as existing loans are repaid or prepaid (this is needed because the assets in the CAST reference pool are amortising, whereas the notes have bullet repayment). This is a departure from the CORE structure, where the collateral and notes are both amortising.

The reference portfolio for the CAST deal is very similar to that for the CORE programme (it represents the ‘Mittelstand’ or German middle-market corporates). It is diversified both geographically across Germany and also across sectors.

For more detail on the CAST 99-1 deal you should read ‘CAST 99-1 A Synthetic CLO’, Deutsche Bank Global Markets Research November 1999.

Exercise

You have been given some brief details below of the GLOBE 2000-1 CLO. You should also read the DB document ‘GLOBE-R 2000-1’, Global Markets Research June 6 2000.

You are required to create a brief sales memo (one page max) in anticipation of phoning an institutional investor. Send your sales memo to : [email protected]

GLOBE 2000-1, synthetic CLO (no SPV), CLN

This was Deutsche Bank’s third synthetic CLO (following on from CAST 1999-1 and Blue Stripe 1999-1).

Like the CAST deal, GLOBE does not have an SPV. The notes in the transaction are credit linked notes that are direct obligations of Deutsche Bank. As in the CAST deal, the AAA notes are collateralised by AAA pfandbriefe, effectively de-linking them from the credit rating of Deutsche Bank. The pfandbriefe are only used to insulate AAA investors from Deutsche Bank credit risk – the credit risk in the reference portfolio is managed through tranching (subordination).

The notes have a bullet repayment structure and a maturity of 5.5 years. The purpose of the deal is for Deutsche Bank to obtain capital relief in order to improve regulatory and economic capital ratios.

The structure is shown below.



GLOBE 2000-1

The tranching is as follows :

Globe Tranche Structure

There is credit leverage in a synthetic CLO because the reference portfolio is much larger than the note issue. The synthetic structure thus allows Deutsche to obtain credit protection on a much larger reference portfolio than would be practicable in the conventional CLO format.



Moreover, the synthetic CLO structure can be used effectively to create bullet structures, whereas conventional CLO’s tend to have amortising notes.

The main difference between GLOBE and the earlier CAST deal relates to the type of assets in the reference pool. In GLOBE the assets are of varying types (syndicated loans, bilateral loans, revolving credits) and denominated in several currencies. Given that the notes are CLN’s and direct obligations of Deutsche Bank AG, there is no specific hedging in the structure [ losses in non-EUR denominated loans are calculated using either the exchange rate valid at the credit event date or the loss determination date, whichever yield the lower loss value in EUR].

Relative Value In CLO’s

The most relevant comparables for balance sheet CLO’s are bank bonds and credit card receivable ABS. CLO’s offer a significant pickup to both asset classes, partly because there is less liquidity in the CLO market compared with the more well-established bond markets.

When synthetics first appeared in the market they tended to trade at a spread to more conventional CLO’s. This is likely to have been linked to how investors perceived the ‘super-senior’ tranche . Some investors feel that there is ambiguity in having a AAA class that is nevertheless subordinate to the super-senior class. They are somewhat cautious in accepting that expected losses on the super-senior and AAA should be the same.

There has also been some confusion amongst AAA-only investors as to whether or not the AAA mezzanine classes are eligible investments.

The premium for synthetic issues over more conventional CLO’s appears to be declining.

☺That’s the end of the session (at last !) You should now :

Understand the rationale for issuing balance sheet CDO’s

Be aware of the credit enhancements for balance sheet CDO’s

Confused ? email Mike Pawley at [email protected]


mailto:[email protected]



Arbitrage CDO’s

Mike Pawley




What will you get out of this material ?

Well, at the end of this session you should :

Understand how cash flow and market value CDO’s are structured

Understand the rationale for issuing CDO’s

Be aware of the credit enhancements for cash flow and market value CDO’s

Be aware of the ratings agencies’ approach to CDO’s

Understand the structures behind single tranche and index CDO’s

Be able to describe CDO2 and CMCDO’s

What you already need to know:

You should already understand :

The basic mechanics of ABS

Balance sheet CLO’s




This session covers arbitrage CDO’s. We start with a reminder about the general CDO product.

The term Collateralised Debt Obligation (CDO) is a general name for investment vehicles backed by loans, bonds, structured products, or various mixtures of all three.

Securities backed by bank loans are called Collateralised Loan Obligations (CLO’s). Securities backed by bonds are known as Collateralised Bond Obligations (CBO’s). More recently, CDO’s have evolved investment vehicles that are backed by structured products (surprisingly enough, called Structured Product CDO’s or ABS CDO’s).

The terminology tends to overlap to some extent. For example, some deals that are called CBO’s may contain loans as well as bonds in the asset pool. Try not to get too carried away with strict definitions – the product terms tend to blur at the boundaries somewhat.

The figure below shows a useful categorisation of the products and also a time tag for each product to give you some idea as to how the product has developed over the last fifteen years or so. The list is not meant to be exhaustive – new innovations are occurring at a rapid pace.

CDO Product Range


CDO’s

CBO’s CLO’s








CMCDO’s2004 on

CDO’s

CBO’s CLO’s








CMCDO’s2004 on


Arbitrage CDO’s

In an arbitrage CDO the issuer captures any spread that might be evident between the pricing of securities (typically ABS, investment grade and high yield) in the secondary capital markets and the yield on investment grade assets that are sold to investors.

Despite the arbitrage that is available to CDO collateral managers, there is also something in this for investors. CDO bonds offer substantially wider spreads across the credit spectrum relative to comparably rated corporate bonds. CDO’s also offer investors diversification, and, typically, access to assets that might not otherwise normally be available in the market.

The figure below shows the outline of a generic CDO.

Conventional CDO Structure

As can be seen from the figure below, the CDO market has to some extent shifted to ABS assets and away from investment grade and high yield assets. ABS are less volatile and more diversified, with stable ratings and low correlation to specific industry risks. If there are problems with an ABS pool there will tend to be slow deterioration in the pool, rather than shock ‘single event’ risks such as Worldcom.



Collateral Pool Types

Arbitrage CDO’s can be either ‘Cash Flow’ or ‘Market Value’. In the cash flow CDO (the more common of the two by number of transactions) the cash flow from the collateral pays the interest on the investors’ notes. Fluctuations in the market value of the collateral are relatively unimportant (except if there is a default) in cash flow CDO’s. The assets tend not to be actively traded – instead, this is a relatively static buy-and-hold type of strategy. Cash flow CDO’s are structured so that investment grade bonds can withstand the worst default experience seen over the past 30 years without suffering losses.

In a market value CDO, by contrast, the asset manager actively trades the collateral to take advantage of perceived changing market conditions. The performance of investors’ securities thus depends on both the market value and credit performance of the collateral pool. Mark-to-market of the collateral occurs regularly, and the asset manager is required to keep an equity cushion between the market value of the assets and the principal amount of investors’ outstanding debt. If this falls below a trigger amount, the collateral manager must sell assets and pay down liabilities. Market value CDO’s are structured so that investment grade bonds can withstand multiples of the worst historic price volatility seen over the past 30 years.

The main difference between CDO’s and most other asset backed securities (eg MBS, CMBS etc) is that the collateral manager in a CDO has much greater scope for actively managing the collateral portfolio. Whilst most classes of ABS tend to have fairly static pools that don’t change greatly, or have limited collateral substitution once they are set up (credit cards being slightly different, given their revolving structure), CDO’s allow for actively managing/trading the portfolio.

The importance of this for investors in market value CDO’s is that the ability of the collateral manager becomes a key consideration in the investment decision.


2000 Full Year

All Other CDOs12%

Trust Preferred

3%Leveraged

Loans25%

HighYield Bonds

24%Emerging

Market1%

Market Value HY4%

Other High Yield6%

InvestmentGrade

9%

ABS/MBSCMBS16%

$55.5 bil.

2004 YTD

All Other CDOs5%

ABS/MBSCMBS64%

Other High Yield2%

Leveraged Loans21%

Trust Preferred

16%

$18.0 bil.


Cash Flow CDO’s

Cash flow arbitrage structures tend to dominate the arbitrage CDO market, representing around 60-70% of the total number of CDO transactions, and about 40-45% by issuance volume.

The structure of a cash flow CDO is fairly well represented by the generic structure in the figure seen in the previous section.

Cash flow CDO’s are structured so that the collateral pool creates enough interest and principal cash flows to cover debt service requirements on the CDO securities.

Cash flow CDO collateral in the US market (the main market) tends to be dominated by ABS. However, several types of assets have been used in the past to create cash flow CDO’s, including G7 high yield bonds, other OECD high yield bonds, emerging market debt, US and non-US bank loans, mezzanine debt and credit derivatives.

The figure below shows the typical life-cycle of a CDO (both cash flow and market value).

Typical CDO Timeline


Reinvestment Period Paydown Period

Call PeriodNon-Call Period

RampUpWareHousing

Pricing/Closing

8-12 Years

Maturity

Event ABS CDO

1. Warehouse Period 2-3 months

2. Pricing/Closing -

3. Ramp Up 2-3 months

4. Reinvestment Period 0 to 5 years

5. Non-Call Period 3 years

6. Expected Final Maturity 7-13 years

7. Legal Final Maturity 30-35 years


During the ramp-up period the asset manager acquires assets until the transaction is fully invested. In the reinvestment or revolving period, interest payments from the collateral are passed through to bondholders, whilst principal repayments are reinvested in new assets. At the end of the reinvestment period principal repayments are passed through to bondholders, beginning with the most senior class, until all outstanding bonds have been repaid.

Asset Criteria

There are a variety of constraints on the composition of the collateral pool that the collateral manager must meet at inception and on an ongoing basis in a cash flow CDO. If the collateral pool fails these criteria during the life of the transaction, any subsequent trades by the asset manager must either maintain or improve the stated portfolio criteria.

There will be a constraint or limit on the type of asset the CDO can hold (eg limits on the amount of emerging market debt as a percentage of the whole collateral pool, or limits on investing in the notes of other CDO’s, etc).

Typically, the collateral pool will have to maintain a maximum weighted average rating score and a minimum diversity score (Moody’s). There will also usually be concentration restrictions for individual obligors and for individual sectors. This means that CDO portfolios tend to be more diversified than the cash high yield market.

There may also be a minimum coupon requirement (to make sure there are adequate interest flows to meet servicing requirements on the CDO securities) and also a requirement that collateral assets mature before the stated final maturity date of the CDO, to reduce risk of shortfalls at maturity.

Cash flow CDO’s do not have a mark-to-market requirement (unlike market value CDO’s). The deal is structured so that, as long as there are no defaults, the cash flows from the asset pool are sufficient to pay interest and repay principal on all classes of bonds over the life of the deal.

In a typical cash flow CDO the senior class represents 65%-75% of the capital structure, and are usually floating rate. They typically have an average life of around 6 to 8 years (legal final around 10-12 years) and are often callable after 3 to 5 years at the option of the equity class holders at the greater of par or some make-whole provision.

The AAA rated senior class gets its rating based on the internal credit enhancement. This comes in the form of subordination of the mezzanine and junior classes to the senior class. It also comes from overcollateralisation and interest coverage triggers (see below). Notice that the amount of credit enhancement to achieve a AAA rating is substantially higher than compared to other ABS asset classes, largely because of the higher default risk of the collateral and the relatively long term of the assets.

Senior bondholders have minimal credit risk as shown by their AAA or AA ratings, but their expected return is of course lower than the underlying collateral.



The structure may have a single A or BBB mezzanine class, or both, as well as a BB or single B tranche. These classes may be either fixed or floating, according to investor demand. The mezzanine class tends to account for 10-20% of a typical cash flow CDO, whilst the most junior rated class may well account for less than 5% of the deal.

The subordinated classes have a leveraged exposure to credit risk, and offer substantially wider spreads than the underlying collateral.

The equity class is often around 10% of the capital structure (a larger equity class may be required if the collateral pool has emerging market securities or distressed debt).

The table below shows the breakeven default rate on a generic cash flow CDO . The AAA class, for example, can withstand annual defaults of 29.4% in the pool over the life of the deal before losses are taken. This is much higher than the worst one-year high yield default rate recorded by Moody’s (10.53%) or the highest 10 year moving average default rate (5.6%). Even the BBB class can withstand anything that history can throw at it. The B1 class may, however, see losses according the Moody’s data. Investors need to weigh up the risk and returns from these different tranches.

Breakeven Default Rates for Generic Cash Flow CDO’s

Class Rating Amount Loss at Default Rate Of :A AAA 70% 29.4%B Aa2 13% 19.4%C Baa2 6% 11.4%D B1 11% 7%Adapted from “The Arbitrage CDO Market”, Deutsche Bank Global Markets Research, March 21 2000

Overcollateralisation

Cash flow CDO’s use overcollateralisation and interest coverage triggers (as well as subordination) to achieve credit enhancement. If these triggers are breached, interest cash flows from the collateral must be diverted from the more junior classes to pay down the senior notes until the coverage tests are met.

Each class tends to have an overcollateralisation level determined by the rating agencies. A generic overcollateralisation structure might be 125% for the senior class, 108% for the mezzanine, and 100% for the lowest rated class.

The senior class O/C ratio equals the sum of the face amount of the outstanding collateral in the pool divided by the outstanding face amount of the senior class:

100/70 = 143%

The mezzanine O/C ratio is the face amount of the collateral divided by the sum of senior and mezzanine bonds :

100/(70+16) = 116%

The table below gives the figures.



Overcollateralisation Table

Start of

Deal After Loss

8 Pay Down Class A

Par Value

Required O/C

Actual O/C

Par Value Actual O/C

Par Value

Actual O/C

A AAA 70 125% 143% 70 131% 64 134%B BBB 16 108% 116% 16 107% 16 108%C BBB 4 100% 111% 4 102% 4 102%Equity NR 10 - 2 2Total 100 92 86Source : Adapted from “The Arbitrage CDO Market”, Deutsche Bank Global Markets Research, March 21 2000

Go to spreadsheet ‘CDOs – Overcollateralisation’ to see the calculations.

At the start of the life of the structure, all of the classes have higher actual O/C’s than required. However, let’s say that there is a loss of 8 in the collateral. The face value of the collateral drops to 92, reducing the equity class from 10 to 2. The mezzanine class actual O/C is now only 107%, below the required level of 108%. Interest flows must therefore be diverted from the subordinate class to pay down first the senior class then the mezzanine class. In our example, once the senior class is reduced to 64 the mezzanine minimum overcollateralisation test is met.

Deferred interest is paid to the class C noteholders out of excess spread once the overcollateralisation tests are met.

Interest Coverage Ratio TestInterest coverage ratios are calculated by dividing the total interest generated by the collateral by the amount of interest needed to repay expenses and service each class of debt plus all classes above it.

The interest coverage ratio test is constructed to ensure that the interest on the collateral pool is sufficient to service the outstanding debt.

Typical minimum interest coverage ratios might be 140% for the senior class, 125% for the mezzanine and 100% for the subordinate class. The figure below shows the priority of interest and principal payments in a typical CDO. Around 50%-75% of the manager’s fee is a senior obligation of the CDO transaction. The balance is subordinated to the mezzanine class and senior to the equity tranche. Generally, the manager holds 10% to 40% of the equity in the transaction, to keep the manager on his toes.

Interest flows from the collateral are paid to bond holders during the reinvestment period, whilst principal cash flows are reinvested in new collateral. If the structure fails



to meet overcollateralisation or interest coverage test then interest flows are diverted from more junior classes to pay down senior bonds until the tests are met (as explained above).

Cascade of Interest Flows

Source : Adapted from “The Arbitrage CDO Market”, Global Markets Research, March 21, 2000

During the principal repayment period, principal cash flows are distributed to bondholders sequentially as the collateral matures.

Cash Flow CDO Ratings

The main areas of analysis for the rating agencies in a cash flow CDO are the credit quality and cash flow of the asset pool, the qualifications of the asset manager and the legal infrastructure.

Special attention is paid to the level of expected losses on the collateral. All of the rating agencies have built up databases on historical default levels that they use to calculate expected loss.

[Expected loss is a function of the probability of default and loss severity if default occurs. eg if probability of default is 5% and the loss rate on default is 70% of par (a recovery rate of 30%) then the expected loss is 5% x 70% = 3.5%].

Loss severity of a particular asset is tied to the seniority of the asset in the issuer’s capital structure. So, senior secured bank debt has an average recovery rate of about 70%, while senior unsecured debt is about 49%. (these are reduced to around 50%-60% and 35%-40% respectively for stress tests).The collateral pool is stressed and credit enhancement levels are established based on the risk characteristics of the pool and the desired rating.



The agencies generally apply a rating factor to each asset in the pool (see below) and then calculate a weighted average rating based on the asset size.

Moody’s Rating Factors for CDO Pool Analysis

Aaa/AAA 1Aa1/AA+ 10Aa2/AA 20Aa3/AA- 40A1/A+ 70A2/A 120A3/A- 180Baa1/BBB+ 260Baa2/BBB 360Baa3/BBB- 610Ba1/BB+ 940Ba2/BB 1350Ba3/BB- 1780B1/B+ 2220B2/B 2720B3/B- 3490CCC+ NACaa/CCC 6500CCC- NA<Ca/<CCC- 10000

Diversity in the collateral pool is another important factor that rating agencies take into account.

Typically, there are restrictions on any one obligor representing more than 2%-3% of the pool, and a maximum of around 8%-12% in any one industry. If the pool is to have higher concentrations the agencies will require higher O/C ratios (ie higher credit enhancement levels).

Moody’s has created a ‘Diversity’ scoring system (see table below).

The diversity score is interpreted as the equivalent of the number of equal sized, independent positions in the pool. If the pool contained 100 bonds of various sizes and a pool diversity score of 40, Moody’s would treat it as 40 equal sized, independent bonds.



Moody’s Diversity Scores

No. Firms in Same Industry Diversity Score1 1.002 1.503 2.004 2.335 2.676 3.007 3.258 3.509 3.7510 4.00

>10 Case-by-Case

The tables below show how the diversity scores of three different pools would be calculated. The first pool has twenty firms, with two firms in each of ten industries. Using the diversity scores above, the total pool diversity would be 15.

Table : Diversity Score for Pool with 20 Firms

Industry Firms Diversity ScoreA 2 1.5B 2 1.5C 2 1.5D 2 1.5E 2 1.5F 2 1.5G 2 1.5H 2 1.5I 2 1.5J 2 1.5

Total Diversity Score 15



The second pool also has twenty firms, but spread unevenly over twelve industries. Nevertheless, it has more or less the same total diversity score as the first pool.

Table : Diversity Score for Pool with 20 FirmsIndustry Firms Diversity ScoreA 4 2.33B 4 2.33C 1 1D 1 1E 1 1F 1 1G 1 1H 1 1I 1 1J 1 1K 2 1.5L 2 1.5

Total Diversity Score 15.66

The final pool has 100 firms spread over twelve industries, with a total diversity score of 42.75.

Table : Diversity Score for Pool with 100 Firms (Typical CDO)Industry Firms Diversity ScoreA 10 4B 8 3.5C 10 4D 7 3.25E 10 4F 4 2.33G 9 3.75H 10 4I 10 4J 5 2.67K 8 3.5L 9 3.75

Total Diversity Score 42.75

A well diversified CDO may have over 100 bonds in the collateral pool and a Moody’s diversity score of 40 or better, as in the simplified pool above.

Clearly, asset manager ability is important. The agencies review the manager’s prior performance record, its trading facilities, back office function, research operation etc.



Synthetic Arbitrage CDO’s (Managed Cashflow Deals)

Fully synthetic arbitrage CDO’s started to enter the market in 2000. These are a natural extension of the early synthetic deals discussed earlier. The innovation here is that the collateral is a pool of single name CDS. Here, investors are effectively selling protection to the sponsor of the transaction, and the sponsor is then selling it on to dealers (see figure below).

Generic Synthetic Arbitrage CDO

An example of the type of structure seen above is Blue-Chip Funding 2001-1 PLC, that also includes a super-senior credit default swap.

You will also tend to see structures that dispense with notes altogether, and instead enter into credit default swaps with investors. A generic structure is shown below. Because there is no exchange of principal the SPV can be done away with. These structured tranches of ‘Portfolio Swaps’ are a very efficient method of transferring risk. Many of these deals tend to have super-senior tranches.

It is therefore possible to create structures where the cost of buying protection in tranched portfolio form is less than the weighted average cost of the individual default swaps that make up the portfolio. Good examples of these types of structures are the Deutsche Bank Repon 2001-14 structure, and the Deutsche Bank Tsar V that is a synthetic structured product deal.



Synthetic CDO with Tranched CDS

Market Value CDO’s

With market value CDO’s, the performance of the structure is based on the mark-to-market performance of the collateral pool. There tends to be greater latitude for actively trading in market value CDO’s than in cash flow CDO’s. Generally, managers are also able to invest in a wider class of assets in market value CDO’s (see the table below).

Possible Market Value CDO Collateral

US domestic high yield bondsG7 high yield bondsOther OECD high yield bondsUS and non-US bank loansCredit DerivativesMezzanine debtConvertiblesPrefsDistressed debtEmerging market debtEquities

The key rationale for market value CDO’s (as with cash flow CDO’s) is arbitrage – the difference between the total return on the collateral and the yield on the CDO securities.



As with cash flow CDO’s, market value CDO’s have minimum diversity requirements, although these can be breached to take advantage of trading opportunities, as long as they are financed with equity (typical limits shown in the table below).

Diversity Limits (Maximums) for Market Value CDO’s

3.5% per issuer15% in any individual industry25% in special situation investments15% in illiquid investments25% in foreign issuers 7.5% in unhedged foreign investments5% in CDO’s

The tranching in a market value CDO tends to differ to that in a cash flow CDO. The most senior class typically involves a revolving loan facility (around 50% of the capital structure) that can be repaid or drawn down at any time. The purpose of this is to give the collateral manager the flexibility to ramp up the portfolio over time or to sell assets and pay down debt to meet O/C requirements.

There also tends to be another senior class that ranks pari passu with the revolver that accounts for around 10%-20% of the capital structure (see figure below). This is sometimes enhanced to AAA through the use of a third party wrap.



Generic Market Value CDO

There are usually three or four additional mezzanine and subordinate classes, carrying ratings from A to B.

The equity class in market value CDO’s tend to be fairly large so that it can provide a cushion against price volatility in the collateral pool.

The table below shows typical advance rates that can be achieved by collateral managers when issuing CDO debt. These are the proportions of the purchase price (and market value on an ongoing basis) that can be financed with the CDO securities.

Moody’s Advance Rates For Market value CDO’s

AAA AA A BBB BB B

Cash 100 100 100 100 100 100

Government Secs 2-10yr 95 95 95 95 95 100

Bank Loans MV>90% 86 90 91 93 94 96

Investment Grade Corp Bonds 85 88 91 94 94 96

Bank Loans MV 80-90% BB- Corp Debt 73 81 87 90 92 92

Bank Loans MV 70-80% BB- Corp Debt 69 75 85 87 89 89

Other Ban Loans CCC+ 52 60 72 79 85 82

Investment Grade CB’s 60 70 80 85 80 90

Non Investment Grade CB’s 52 64 76 81 78 88

Equity, Illiquid Debt 40 50 71 78 78 85



The advance rates mentioned above have been derived from stress testing over a very wide range of economic scenarios, taking into account changes in interest rates, changes in default rates, changes in correlation among asset classes etc.

Go to spreadsheet ‘CDOs – MV’ to see the calculations.

Key tests in the market value structure are O/C and minimum net worth.

The O/C test is driven by the market value of the collateral. The manager must maintain a minimum ratio of collateral market value relative to the par amount of debt outstanding. If a trigger is breached it must typically be solved within ten days or so (by selling assets and paying down debt). Otherwise, the entire collateral portfolio may be subject to liquidation (to ensure that senior holders are repaid before the market value of the pool deteriorates too much).

Minimum net worth ratios are calculated as the market value of the equity divided by the amount of paid-in capital or original equity (where the market value of equity is the market value of assets less the par amount of debt). The minimum net worth ratio might vary from 60% for the senior class to 30% for the subordinated class. If there is a breach, the senior-most debt holders can decide whether or not to start liquidating the portfolio.



Single Tranche CDO’s

Investment banks have honed their credit and correlation risk management techniques to the extent that they now offer single tranche arbitrage cash flow CDO’s.

CDO’s have four types of risk : default, spread, recovery and correlation. By buying protection on a tranche and simultaneously hedging by selling protection on the individual names in the portfolio, default, spread and recovery risk can be managed to a degree. The trader is left with correlation risk.

Default correlation has a significant impact on the shape of the loss distribution. The shape of the loss distribution, in turn, drives the pricing of the CDO tranches. The figure below shows the loss distribution of a pool of assets given low correlation (1%). With low correlation, the assets are more or less independent. There is a low probability of large losses and also a low probability of zero losses. The spread on the senior tranche will tend to be small, the spread on the equity tranche will tend to be large.

What if we increase the correlation ? The resulting loss distribution for the portfolio given a 45% correlation is shown below. With this ‘medium’ correlation, the assets are more likely to default together and the distribution of the right hand tail is longer. This means more risk in the senior tranche and hence a larger senior spread than in the previous example. The probability of zero losses has also increased, so the equity is less risky and the equity spread consequently smaller than before.


Loss Distribution for 1% Correlation

0%

2%

4%

6%

8%

10%

12%

14%

16%

18%

20%

0% 4% 8% 12% 16% 20% 24% 28% 32% 36% 40% 44% 48%

Portfolio Loss (%)

Pro

bab

ilit

y


0.00%

5.00%

10.00%

15.00%

20.00%

25.00%

30.00%

35.00%

0% 4% 8% 12% 16% 20% 24% 28% 32% 36% 40% 44% 48%

Portfolio Loss (%)

Pro

bab

ilit

y


Finally, for high correlation, as in the figure below, the portfolio behaves like one asset with no diversification. The assets either all default or all survive, with a significant probability that there will be a large number of losses, so the spread required to hold the senior tranche is relatively large. The probability of zero losses is very high so the equity spread is relatively small.

So, to repeat (as it’s so crucial), here’s the impact of correlation :

Low correlation, assets more or less independent, low probability of large losses, low probability of zero losses – spread on senior tranche is small, spread on equity tranche is large.

Medium correlation, assets more likely to default together, distribution of tail longer – more risk in senior tranche, larger senior spread than before; probability of zero losses has increased, equity less risky and equity spread smaller than before.

High correlation, portfolio behaves like one asset, no diversification, either all default or all survive, probability of large number of losses is significant, spread required to hold senior tranche is relatively large; probability of zero losses is very high so equity spread is small.

The terminology here can be confusing :

Equity investors are buyers of correlation (long correlation) – increases in correlation over time make the equity tranche less risky (see the figure below).


Correlation and Tranche Spreads

0.00%

5.00%

10.00%

15.00%

20.00%

25.00%

10% 20% 30% 40% 50% 60% 70% 80%

Default Correlation

Sp

read

s (%

)

Equity

Mezzanine

Senior


0.00%

10.00%

20.00%

30.00%

40.00%

50.00%

60.00%

70.00%

80.00%

0% 4% 8% 12% 16% 20% 24% 28% 32% 36% 40% 44% 48%

Portfolio Loss (%)

Pro

bab

ilit

y


Senior investors are sellers of correlation (short correlation) – increases in correlation over time make the senior tranche more risky.

Mezzanine investors are sellers of correlation (short correlation) – increases in correlation over time make the mezzanine tranche more risky, although the sensitivity is not large and will vary with the particular structure of the deal.

Credit Select

Deutsche Bank offers notes linked to customised synthetic credit portfolios under the Credit Select name. These are CDO’s selected and controlled by the investor (or their manager). The advantages of Credit Select are : flexibility over the initial portfolio composition and changes in the portfolio (simple rules apply); transparent pricing for the initial portfolio and switches; small, private transactions are possible; rapid execution can take place (DB underwrites the remaining capital structure); standardised maturity (5 year).

The figures below show that Deutsche Bank can sell any part of the capital structure to an investor and hedge the rest of the risk.

Credit Select – Investor Buys First Loss Notes


Equity

A

AA

AAA

Super Senior

DB Underwrites Risk on Senior 96% of Structure (100 Investment Grade Credit Swaps, each 10m)

Investor Takes 40m First Loss Notes

0%

4%

100%

Equity

A

AA

AAA

Super Senior


Investor Takes 40m First Loss Notes

0%

4%

100%


Credit Select – Investor Buys BBB Mezzanine

TAPAS

TAPAS is Deutsche Bank’s platform for investor-driven CDO of ABS transactions. It is similar to Credit Select. TAPAS offers investors a number of advantages over traditional CDO of ABS structures : the pool is made up of high quality (double-A and triple-A) rated ABS securities, as opposed to BBB; short bullet maturities (5-7Y), without extension risk or prepayment risk, as opposed to amortising collateral and 30-35 Y legal finals; the client takes exposure to pure default risk – eliminating interest rate, FX and extension risk; guaranteed execution since Deutsche Bank will underwrite all the remaining parts of the capital structure.

The table below shows how a Tapas deal might look.

TAPAS Typical Structure

Minimum Number of Assets 35Maximum Concentration of each Asset 3.50%Diversity Score 18

Minimum Rating AA- by S&PAverage Rating AA / AA+ by S&P Rating Composition Minimum 65% AAA

Max Percentage of each Collateral Class 25.00%Collateral Classes : Consumer ABS

Commercial ABSCDOsCMBS


Equity

BBB

A

AA

Super Senior


Investor Takes 10m BBB Notes

AAA

DB Underwrites 30m Equity (First Loss)0%

3%

100%

4%

Equity

BBB

A

AA

Super Senior


Investor Takes 10m BBB Notes

AAA

DB Underwrites 30m Equity (First Loss)0%

3%

100%

4%


RMBSIndex CDO’s

Index CDO’s are similar to synthetic investment grade CDO’s, however, there is an important innovation : Investors can buy and sell protection on tranches. Index CDO’s have thus paved the way for standardised trading of correlation.

Unfortunately, index CDO’s can again cause confusion over terminology. Here is a guide :

Buy Tranche Sell TrancheLong Credit Short CreditSell Protection Buy ProtectionTake Risk Sell Risk

The Dow Jones Indices have become the de facto indicators for the synthetic credit default swap market. The DJ indices represent the merger of two leading indices in June 2004 (iBoxx and DJ TRAC-X). Here are (some) of the indices available :

CDX.NA.IG 125 US namesiTraxx Europe 125 namesiTraxx Asia (ex Japan) 30 names (plus sub-indices for Korea, Greater China and other countries) The Dow Jones iTraxx Indices have become the de facto indictor for the synthetic credit default swap market. Represents the merger of two leading indices in June 2004 (iBoxx and DJ TRAC-X).iTraxx Japan 50 namesiTraxx Australia 25 names

The figure below shows the capital structures of two of the main indices :

Dow Jones Indices Capital Structures


Equity (0-3%)

BBB (3-7%)

AAA (7-10%)

Jnr SS (10-15%)

Super Senior (15-30%)

US

CDX. NA Investment Grade Index

Cash Flows

Losses

Equity (0-3%)

BBB (3-6%)

AAA (6-9%)

Jnr SS (9-12%)


Europe

DJiTraxx

Europe 5y

Cash Flows

Losses

Equity (0-3%)

BBB (3-7%)

AAA (7-10%)

Jnr SS (10-15%)


US


Cash Flows

Losses

Equity (0-3%)

BBB (3-7%)

AAA (7-10%)

Jnr SS (10-15%)


US


Cash Flows

Losses

Equity (0-3%)

BBB (3-6%)

AAA (6-9%)

Jnr SS (9-12%)


Europe

DJiTraxx

Europe 5y

Cash Flows

Losses

Equity (0-3%)

BBB (3-6%)

AAA (6-9%)

Jnr SS (9-12%)


Europe

DJiTraxx

Europe 5y

Cash Flows

Losses


For traders and investors the tranche delta is an important concept for hedging purposes. The tranche Delta with respect to the index (sensitivity of tranche to change in underlying index spread) is :

Change in Tranche MTM Change in Index Spread

Eg delta of -70 means that the change in tranche MTM is -0.70% (ie -70 x 0.01%) for a one basis point shift in the index spread from say, 42 to 43 bp. So, another way to think of this is that -0.70% is the tranche PV01.

Tranche PV01 wrt Index = Tranche Delta wrt Index x 0.01%

How does hedging work ? Here’s one way to hedge :

Sell protection on a tranche – buy protection on the underlying index to hedgeBuy protection on a tranche – sell protection on the underlying index to hedge

Hedge Ratio : Tranche PV01 wrt Index Index PV01

However, there are problems with delta hedging using the index. If the hedger believes that general spread changes need to be hedged then using the index is fine – but if the hedger believes some credits may move independently then hedge exposures to individual credits. This is because the index sensitivity is the same for all credits because they are equally weighted, but the tranche sensitivity varies across credits. Delta hedging also results in negative carry for all but the equity tranche.

Another important factor for tranche hedgers is gamma risk. Tranche gamma is the sensitivity of the tranche delta to changes in the underlying index spread :

Change in Tranche Index DeltaChange in Index Spread

Equity investors are said to be long gammaSenior investors are short GammaMezzanine investors are Short then long gamma

Tranche delta with respect to crediti

Tranche Delta wrt Crediti (sensitivity of tranche to change in single name (crediti) spread :

Change in Tranche MTMChange in Credit Spread of Crediti

Tranche PV01 wrt Crediti = Tranche (Crediti) Delta x 0.01%



Hedge Ratio :

Tranche PV01 wrt Crediti Crediti PV01

In practice, hedgers often delta hedge using individual credits. Here, a tranche is hedged (in theory) by buying protection on delta hedge notionals on all credits in the pool. Carry is again negative for all but the equity tranche, and rebalancing is very expensive and resource intensive. In practice, many many hedgers use a reverse cherry picking strategy whereby they only hedge the riskiest names. The problem, of course, is how to identify the riskiest names – by using spread or asset volatilities perhaps ?

There is also still gamma risk :

Sensitivity of tranche delta to changes in underlying index spread

Change in Tranche Delta wrt Crediti

Change in Index Spread of Crediti

Equity investors are said to be short gamma.Senior investors are long gamma.Mezzanine investors are long Gamma.

CDO2

The market has evolved to the extent that CDO of CDO’s are possible (or, in the terminology, CDO Suared or CDO2). The figure below shows that the CDO master pool is made up of the mezzanine tranches of several CDO’s.

CDO2 Structure

Global Markets Training 50Underlying Pool

Asset NAsset (N-1)Asset 3Asset 2Asset 1

EquityEquityEquityEquity

MezzanineMezzanineMaster PoolMezzanineMezzanine

seniorseniorseniorsenior

Senior/SuperSenior/SuperSenior/SuperSenior/Super

CDO Pool NCDO Pool (N-1)CDO Pool 2CDO Pool 1

Equity

Mezz.(Low)

Mezz. (High)

Senior

CDO Squared

Underlying Pool

Asset NAsset (N-1)Asset 3Asset 2Asset 1

EquityEquityEquityEquity

MezzanineMezzanineMaster PoolMezzanineMezzanine

seniorseniorseniorsenior

Senior/SuperSenior/SuperSenior/SuperSenior/Super

CDO Pool NCDO Pool (N-1)CDO Pool 2CDO Pool 1

Equity

Mezz.(Low)

Mezz. (High)

Senior

CDO Squared


There is typically high diversification in these deals, with an underlying pool of usually around 300 names and a diversity score of 200-250 as opposed to 75-100 for vanilla deals. Investors tend to buy CDO2 tranches to express macro views and systemic risks. This is because the number of defaults needed to hit mezzanine tranches is typically fairly high, and would normally result from default contagion ie large shocks to the global economy. The concept of ‘overlap’ or ‘double leverage’ is crucial to CDO2. It means the existence of the same credit in several constituent CDO’s underlying the CDO2. The overlap effect means the investor receives higher tranche spreads for the same rating. However, overlap also increases the risk to systemic risk.

The figures below should prove interesting….They show that overlap tends to have the same impact on CDO loss distributions as correlation. The lower the overlap, the lower the correlation , the higher the overlap, the higher the correlation. So, equity returns will fall as overlap rises whereas mezzanine and senior returns will tend to rise.


Overlap and Tranche Spreads

0.00%

75

150

225

300

375

- Low - HighOverlap

Sp

read

s (%

)

EquityMezzanine*Senior

0%

2%

4%

6%

8%

10%

12%

14%

16%

18%

20%

0% 4% 8% 12% 16% 20% 24% 28% 32% 36% 40% 44% 48%

Portfolio Loss (%)

Pro

bab

ility

0.00%

5.00%

10.00%

15.00%

20.00%

25.00%

30.00%

35.00%

0% 4% 8% 12% 16% 20% 24% 28% 32% 36% 40% 44% 48%

Portfolio Loss (%)

Pro

bab

ility

0.00%

10.00%

20.00%

30.00%

40.00%

50.00%

60.00%

70.00%

80.00%

0% 4% 8% 12% 16% 20% 24% 28% 32% 36% 40% 44% 48%

Portfolio Loss (%)

Pro

bab

ility

Medium correlation, medium overlap, lower equity returns, higher senior returns

High correlation, high overlap, lowest equity returns, highest senior returns

Low correlation, low overlap, high equity returns, low senior returns

*Depends on Tranche Structure


0.00%

75

150

225

300

375

- Low - HighOverlap

Sp

read

s (%

)



0.00%

75

150

225

300

375

- Low - HighOverlap

Sp

read

s (%

)


0%

2%

4%

6%

8%

10%

12%

14%

16%

18%

20%

0% 4% 8% 12% 16% 20% 24% 28% 32% 36% 40% 44% 48%

Portfolio Loss (%)

Pro

bab

ility

0%

2%

4%

6%

8%

10%

12%

14%

16%

18%

20%

0% 4% 8% 12% 16% 20% 24% 28% 32% 36% 40% 44% 48%

Portfolio Loss (%)

Pro

bab

ility

0.00%

5.00%

10.00%

15.00%

20.00%

25.00%

30.00%

35.00%

0% 4% 8% 12% 16% 20% 24% 28% 32% 36% 40% 44% 48%

Portfolio Loss (%)

Pro

bab

ility

0.00%

5.00%

10.00%

15.00%

20.00%

25.00%

30.00%

35.00%

0% 4% 8% 12% 16% 20% 24% 28% 32% 36% 40% 44% 48%

Portfolio Loss (%)

Pro

bab

ility

0.00%

10.00%

20.00%

30.00%

40.00%

50.00%

60.00%

70.00%

80.00%

0% 4% 8% 12% 16% 20% 24% 28% 32% 36% 40% 44% 48%

Portfolio Loss (%)

Pro

bab

ility

0.00%

10.00%

20.00%

30.00%

40.00%

50.00%

60.00%

70.00%

80.00%

0% 4% 8% 12% 16% 20% 24% 28% 32% 36% 40% 44% 48%

Portfolio Loss (%)

Pro

bab

ility

Medium correlation, medium overlap, lower equity returns, higher senior returns

High correlation, high overlap, lowest equity returns, highest senior returns

Low correlation, low overlap, high equity returns, low senior returns

*Depends on Tranche Structure


CMCDO’s

The premium of a CMCDS in the first year is X% of the 5 year spot CDS spread, subsequent premiums are reset at a pre-determined frequency (quarterly, semi-annually), to X% of the then spot 5 year spread.

X% is known as the ‘participation rate’ and stays constant.

Selling CMCDS protection provides spread income plus the upside to spread widening/curve steepening (whilst still taking default risk ie long credit) – the best analogy is fixed bonds versus FRN’s, CDS versus CMCDS.

CMCDS should be used when spreads are tight, there are flat curves and vol is lowie Now !

CMCDS generally provide higher spread income than short-dated risk and lower MTM volatility compared to a plain vanilla CDS. Note that the maturity of the contract and the maturity of the CMCDS reference rate need not be the same.

CMCDO’s have an underlying reference pool of CMCDS, and the notes provide investors with a floating credit spread product to allow them to express spread widening views. Investors get higher spreads as CMCDS spreads widen, but greater structural complexity and lower liquidity. For mezzanine and senior tranches any spread widening is likely to outweigh the reduction in MTM due to increased default risk.

Recommended Reading :

‘The Arbitrage CDO Market’, DB Global Markets Research, March 2000‘Structured Product CDO’s’, DB Global Markets Research, March 2001‘Understanding Synthetic Structured Product CDO’s - CDObserver Special Report’, DB Global Markets Research, June 2002‘A Practical Framework for CDO Debt Valuation - CDObserver Special Report’, DB Global Markets Research, October, 2002‘2003 ISDA Credit Definitions’ DB Global Markets Research, June 2003‘New Frontiers in Correlation : CDO Squared Products’ DB Global Markets Research, May 2004Structured Products : Implementing our Views (CMCDO) Global Markets Research, March 2004‘Trading Tranched Index Products : First Steps’ Global Markets Research, July 2004‘Trading Tranched Index Products : Our Deltas Explained’ Global Markets Research, August 2004‘ABC of CDO’ Credit Magazine (in association with DB), 2004



☺

That’s the end of the session (at last !) You should now :

Understand how cash flow and market value CDO’s are structured

Understand the rationale for issuing CDO’s

Be aware of the credit enhancements for cash flow and market value CDO’s

Be aware of the ratings agencies’ approach to CDO’s

Understand the structures behind single tranche and index CDO’s

Be able to describe CDO2 and CMCDO’s

Confused ? email Mike Pawley at [email protected]

This publication is for internal use only by Deutsche Bank Global Markets employees. The

material (including formulae and spreadsheets) is provided for education purposes only and

should under no circumstances be used for client pricing. Examples, case studies, exercises and

solutions may use simplifying assumptions that do not apply in practice, and may differ from

Deutsche Bank proprietary models actually used. The publication is provided to you solely for

information purposes and is not intended as an offer or solicitation for the purchase or sale of

any financial instrument or product. The information contained herein has been obtained from

sources believed to be reliable, but is not necessarily complete and its accuracy cannot be

guaranteed.


Cdos

Documents

balance sheet cdos

abs cdos

structured product cdos

synthetic balance sheet

index cdos

synthetic arbitrage

single tranche cdos

arbitrage cdos page