Friday, March 09, 2007

I'm shocked, shocked to find that subprime is in this CDO!

This story from the Wall Street Journal is tough to swallow. The gist is that some European CDO buyers are surprised to find out that their CDO of ABS has significant sub-prime exposure. I don't know whether to laugh or to cry. I laugh because these people must be like the Keystone Kops of portfolio managers. I cry because these people undoubtedly run bigger portfolios than I do.

1. If you've never seen marketing docs for a CDO, I'm here to tell you the allowable securities are spelled out quite clearly. How much you wanna bet that if we pull the offering docs from the CDOs in question there is a page titled Expected Portfolio Mix? And on that page there is a pretty pie chart that has a nice big blue slice labeled B/C Home Equity?

2. I don't have any stats on this, but the overwhelming majority of ABS CDO's I've seen have a big chunk of subprime residential MBS. In order to construct a CDO, you need higher-yielding bonds. In the ABS world, sub-prime residential gives you both the yield and the issuance volume needed to do a large CDO. There are ABS CDOs done with speciality sectors, but I've got to believe that the majority include sub-prime MBS.

So given #1 and #2, in order for an investor to be surprised there was sub-prime in his CDO, s/he did not read the marketing documents, and was unaware of the common structure of securities s/he was buying. In other words, they didn't understand the sector, nor did they bother reading anything to learn more.

On top of this, a CDO investment is all about the riskiest part of the collateral. For those who don't deal in the CDO market, here is a quick overview on how CDO's work.

The CDO sells debt to the public and uses the proceeds to buy a portfolio of bonds. CDO debt is sold in various tranches. Each tranche gets paid sequentially from the most senior to the most junior. In a simple CDO structure, there might be 4 tranches. Class A gets paid interest and principal first, then Class B, then Class C, then Class D. If there is anything left over, that amount passes through to an equity holder. You can see that if the collateral portfolio starts to take credit losses, there might not be enough money to pay the more junior classes.

Usually the higher quality the portfolio, the smaller the junior classes are as a percentage of the whole structure. For example, if the collateral portfolio has an average rating of Ba2, the Senior tranche might be 65% of the total structure. However, if the average rating is A3, the Senior tranche is probably something like 90%. The percentage of the structure which is junior to your bonds is called subordination. Obviously the more subordination you have, the safer your bond is.

Most ABS deals have very little subordination. If the most Senior bond has only 10% subordination, than the Class C might only have 2-3%. Let's assume (as is common) that the Class C tranche was rated A when the deal was sold, and had 3% subordination.

So we know that if the deal suffers 3% losses (not defaults, but losses) immediately, then your A-rated bond now has zero subordination, and every subsequent loss will hit your tranche directly. In real life, losses don't happen on day 1, so I'm being overly simplistic by using 3%. But you get the picture. Losses don't have to be that large in order for investment-grade bonds to start taking hits to principal.

Sticking with the 3% number to keep the math simple, let's assume that subprime home equity has a 50% recovery rate. Furthermore, let's assume the collateral in the CDO is 80% prime MBS and 20% subprime. The investor might say "Hey, I only have 20% subprime exposure." Wrong. Because you only have 3% subordination. If recovery is 50%, then 6% defaults eliminate all your subordination. The 80% in prime MBS doesn't protect you if the 20% in subprime defaults at a high rate.

Now imagine you've bought 10 A-rated CDO deals which have a mix of credit cards, auto loans, student loans, and other ABS. But each of them has around 20% in subprime MBS. Well guess what? If subprime defaults at a high rate, its possible that all 10 of your CDO's will wind up defaulting at the same time. Diversified portfolio? Hardly. Because a CDO investment is all about the riskiest part of the collateral. You could own 1,000 different deals, if the riskiest part of each deal is the same, you have no diversity.

So these investors did not understand the nature of the structure they were buying, did not know what the underlying collateral was, and did not read the marketing documents to try to find out.

Now, please excuse me while I call my broker to short European bank stocks.


Anonymous said...

Certainly agree about these putative portfolio managers; it appears they may be putting a lot at risk, wonder if it includes their own money too.

But thanks for the simplified explanation of CDO structure (which I didn't previously understand). Is this structure typical of MBS or securitized debt generally? How do tranches get the sorts of security ratings they do -- e.g., A or AA -- and still have potential hits to principal potentially that close?

Accrued Interest said...

CDO's are going to be very much in the headlines if this subprime problem becomes bigger and bigger. So I think I'll post a better CDO primer soon.

Anonymous said...

Could you work through a more detailed explanation of the maths of your example.

Anonymous said...

Investors that bought US ABS CDOs knew what they were getting into.

The investement banks did not obviously do much to stress the risk, they just left it there.

The US investement banking model of investment banks originatig mortgages and securitising them (the SISO model, or "sh*t in sh*t out") is the wrong model for the credit markets. Sadly, it is being also adopted in Europe.

At heart lies also a big fundamental problem which is that the rating agencies don't have any hurt money at risk in the deals they rate.

Wait 3 years and you will see similars problem in Europe.