Monday, February 18, 2008

Misleading Chart of the Day, CDS Edition

(Felix Salmon) This graphic comes from Gretchen Morgenson's front-pager in the NYT yesterday. I'm not going to try to reproduce it here, because my column width isn't big enough to really see what's going on. But suffice to say that it shows the market in credit default swaps, at $45.5 trillion, dwarfing the markets in US stocks ($21.9 trillion), mortgage securities ($7.1 trillion), and US Treasuries ($4.4 trillion).

Morgenson's article makes it clear that it's reasonable to directly compare market sizes like this. Indeed, she refers to CDSs as "securities" in the third paragraph of her piece:

The market for these securities is enormous. Since 2000, it has ballooned from $900 billion to more than $45.5 trillion -- roughly twice the size of the entire United States stock market.

But of course a credit default swap is not a security, it's a derivative. The $45.5 trillion is a notional amount; the size of the stock market is a hard valuation. There's an enormous difference.

Morgenson is right that there are problems in the CDS market. But she over-eggs her pudding so much that it's very hard to separate the good points from the bad. And when she leads with the statement that credit default swaps are securities, it's hard to take the rest of the article seriously.

There's another chart in the article, too, which explains that one counterparty to a CDS transaction can "assign" his liability to a third party who may have to be "tracked down" and who "may or may not be in a position to pay".

The irony here is that the problem of assigning CDS liabilities to third parties is actually relatively small for exactly the same reason that the nominal size of the CDS market is so big. Let's say a trader at a hedge fund or an insurance company has sold credit protection to a client. In order to balance out his books, he can do one of two things: either assign that contract to a third party, or simply buy the same amount of credit protection from someone else. The way the CDS market works, he will almost always do the latter rather than the former.

As a result, there are lots of equal-and-opposite contracts being held on the books of all manner of banks and hedge funds and insurance companies. When people tot up big numbers like $45.5 trillion, they're adding all those contracts up together as though they don't cancel each other out, when the whole point of writing them was precisely that they offset existing positions.

Of course, that doesn't mean counterparty risk has gone away - far from it. But it does mean that the problem of tracking down your counterparty is probably smaller, in practice, than Morgenson makes it out to be: in the vast majority of cases, your counterparty will be whoever you entered into the contract with in the first place.

Morgen also uses up a lot of space talking about what happens in an event of default, when the protection seller has to pay up. But it's not clear if she's worried that protection sellers will have to pay too much, or whether she's worried that protection sellers will have to pay too little. At first, it seems like the former:

A bank that has bought protection to cover its corporate bond exposure thinks it is hedged and therefore does not write off paper losses it may incur on those bond holdings. If the party who sold the insurance cannot pay on its claim in the event of a default, however, the bank's losses would have to be reflected on its books.

In this case, a bondholder buys protection against default, but never gets paid by the protection seller, who can't afford to pay out on all the claims he's written. Later on, though, the problem is different:

When Delphi, the auto parts maker, filed for bankruptcy in October 2005, the credit default swaps on the company's debt exceeded the value of underlying bonds tenfold. Buyers of credit insurance scrambled to buy the bonds, driving up their price to around 70 cents on the dollar, a startlingly high value for defaulted debt.
Market participants worked out an auction system where settlements of Delphi contracts could be made even if the bonds could not be physically delivered. This arrangement was done at just over 36 cents on the dollar; so buyers of protection on Delphi who did not have the bonds received $366.25 for every $1,000 in coverage they had bought. Had they been valuing their Delphi insurance coverage at $1,000 per bond, they would have had to write off that position by $633.75 per $1,000 bond.

In this case, a bondholder buys protection against default, but only gets $366.25 for every $1,000 of bonds he insured. If he thought that protection was worth $1,000, then yes, there will have to be write-offs. But no one values credit default swaps as though recovery value is zero. And if the bonds are trading at 70 cents on the dollar, then it's worth pointing out that a bondholder with $1,000 face value of bonds, plus an insurance contract on those bonds, now has bonds worth $700 and an insurance contract worth $366.35, for a total of $1,066.25. Which doesn't sound like too much of a loss to me.

No comments: