Statistical & Financial Consulting by Stanford PhD
Home Page

Collateralized Debt Obligation (CDO) is a collection of securities called tranches. Tranches receive cashflows dependent on the number and severity of defaults in a specific "collateral pool". The pool is called the reference portfolio and is compiled of debt obligations of the same kind: either bonds or loans, or other credit instruments. Tranches are prioritized in the sense of some tranches receiving the cashflows before the others. This is how the whole structure works.

The underwriter of the CDO, typically an investment bank, decides which assets go into the reference portfolio. There are three options:

  • real cashflow generating assets, e.g. bonds or mortgages (cash CDO),

  • the protection seller positions in credit default swaps written on corporate or sovereign entities, or other credit derivatives (synthetic CDO),

  • contracts replicating cashflows on a given pool of defaultable securities but not providing any ownership rights ("truly synthetic" CDO).
In a cash CDO the reference assets can be corporate or sovereign bonds (collateralized bond obligation), loans (collateralized loan obligation), mortgages (collateralized mortgage obligation), real estate investment trusts and so on. In a synthetic CDO corporate credit default swaps are the most popular collateral. There are also so-called "CDO squared". These are CDOs where the reference portfolio is composed of tranches from other CDOs (usually the tranches that nobody wanted and the underwriter had to dispose of one way or another).

A special purpose vehicle (SPV) is set up to buy the reference portfolio and redistribute its cashflows. The special purpose vehicle is an entity legally different from the CDO underwriter. SPV slices the cashflows into tranches, which are set up as separate financial securities sold to different clients, generally speaking. Each tranche entitles the holder to a fixed fraction of the cashflows from the reference portfolio. The cashflows may come at expiration of the CDO or earlier. They may come as pure notional or as notional plus coupon. The tranches do not have equal rights and are subject to strict hierarchy. If the reference portfolio suffers through defaults, then the losses are absorbed by the more junior tranches before the more senior tranches. The most senior tranche is called "super-senior". It is relatively low-risk and is entitled to the lowest coupon. The most junior tranche is referred to as the "equity tranche", since the risk profile of the equity tranche holders is similar to that of stock equity holders. It is customary to slice a CDO into many tranches. The CDO may be composed of the super-senior tranche, senior tranche, senior mezzanine tranche, mezzanine tranche, junior mezzanine tranche and equity tranche. But it does not have to be that way. Any specific deal may implement a "thinner" or "thicker" slicing scheme.


Traders distinguish between standardized CDOs, where the reference pool and contractual provisions are standard, and bespoke CDOs, where all the terms and conditions are customized for specific clients. A good example of standardized CDOs are tranches on liquid credit indexes, e.g. CDX Investment Grade, CDX High Yield and iTraxx. The credit indexes and tranche specifications are defined and monitored by Markit Group Limited. For these Markit tranches there is a liquid over-the-counter market. Big banks offer market making with relatively tight bid-ask spreads for 3 tenors out of 4 for each underlying index. The lower part of capital structure (tranches with lower seniority) tends to enjoy higher volumes and tighter bid-ask spreads.

Standardized and bespoke CDOs are traded over the counter (OTC). Only a subset of big prime brokers will allow you to trade CDOs and for that luxury you will have to pay the yearly fee exceeding $300,000 or trade hundreds of millions of notional per year.

Whenever a big bank makes markets on a liquid CDX IG tranche, there is somebody on the other side of the trade. Oftentimes this is a hedge fund and oftentimes the trade is part of a systematic strategy assuming a relatively frequent rebalancing. Hedge funds do try to introduce semi-systematic, semi-algorithmic, semi-quantitative ideas that have worked in futures and stocks markets into the world of structured credit. Still, the effort has been feeble from what is known to the public. Most credit funds adopt a "buy and hold" policy. This is when you think that you can price a CDO better than the market maker, so you enter the trade at a favorable level, hedge the hedgeable risks and warehouse the rest. The approach was very sharply depicted in Michael Lewis' book referenced below.


Suppose that each of the 20 assets in the reference portfolio has a 3% probability of defaulting during the life of the CDO. If defaults are independent, then the likelihood of the equity tranche experiencing no losses is 54.38%. On the other hand, if the defaults are perfectly correlated, the likelihood of the equity tranche experiencing no losses is 97%. So correlation is beneficial for the equity tranche. We can see how CDO pricing can be sensitive to the correlation of the reference assets.

The introduction of copulas into structured credit by David Li offered a simple way of capturing default correlation. It allowed to streamline the pricing of CDO tranches and ease the development of corresponding risk reports. The copula models may have dozens of flaws but they ensure relatively simple calibration of parameters, speed of calculations and low memory usage. The practical aspects of copula models make them the default method for most trading desks on the buy and sell side.


Mahadevan, S., Mulfeldt, A., & Naraparaju, P. (2011). Handbook of Credit Derivatives and Structured Credit Strategies: Credit Derivatives Insights (5th ed). Morgan Stanley.

Choudhry, M. (2010). Structured Credit Products: Credit Derivatives and Synthetic Securitisation (2nd ed). Wiley.

Duffie, D., & Singleton, K. (2003). Credit Risk: Pricing, Measurement, and Management. Princeton University Press.

Duffie, D. (2001). Dynamic Asset Pricing Theory (3rd ed). Princeton University Press.

Duffie, D., & Singleton, K. (1999). Simulating Correlated Defaults. Working Paper, Stanford University.

Jarrow, R. A., & Yu, F. (2001). Counterparty Risk and the Pricing of Defaultable Securities. Journal of Finance 56, pp. 1765-1800.

Schönbucher, P., & Schubert, D. (2001). Copula-Dependent Defaults in Intensity Models. Working Paper, Bonn University.

Meneguzzo, D., & Vecchiato, W. (2004). Copula Sensitivity in Collateralized Debt Obligations and Basket Default Swaps. Journal of Futures Markets 24(1): 37 - 70.

Li, D. X. (2000). On Default Correlation: A Copula Function Approach. Journal of Fixed Income. 9 (4): 43–54.

Markit (2008). Markit Credit Indices, a Primer. Markit Group Limited.

Tavakoli, J. M. (2008). Structured Finance and Collateralized Debt Obligations: New Developments in Cash and Synthetic Securitization (2nd ed). Wiley, Hoboken, New Jersey.

Lewis, M. (2011). The Big Short: Inside the Doomsday Machine. W. W. Norton & Company.