When Mortgage Derivatives Get Included in CDOs

Antony Currie of Breaking Views, who's been following the mortgage mess very closely, emails me with some color about the degree to which derivatives – credit default swaps, or CDSs – are a part of the CDOs that everybody seems to be so worried about these days.

Many CDOs, including subprime-heavy CDOs, did in fact use mortgage CDS, he says, either exclusively or in conjunction with the actual bonds. Part of the reason is that they wanted the riskiest tranches of the CDOs, and there weren't enough of those to go around:

Sure, $483bn of mortgage bonds were issued. But a lot of CDOs (dubbed mezzanine structured finance CDOs) only wanted to buy the lower-rated slices of MBS – from BB to BBB+. These tranches combined only accounted for 5%- 7.5% of an MBS deal – or $24bn to $36bn. And not all of it went to CDOs (though probably most of it).
So, single-name mortgage CDS, when they were liquid, were lapped up by CDOs. And not just because someone else already owned the bonds. A CDO manager might have liked the risk profile of a particular MBS deal so much that he or she would have bought exposure to it for lots more CDOs he or she managed.
As one CDO manager put it to me, a single, BBB-rated bond of $10m could have perhaps $300m of notional outstandings through the CDS market.

This is beginning to make a bit more sense – if a CDO manager wanted a lot of exposure to the riskiest tranche of a mortgage-backed bond, then even if that tranche was relatively small he could get the exposure he wanted by writing default protection on it.

But it's worth remembering that a CDS contract, like any derivative, is a zero-sum game: where there's a loser, there's a winner. This is why I think it's very unhelpful to think of CDOs themselves as derivatives, especially when they only invest in bonds and loans. Everybody can lose money by investing in a bad debt instrument, but if somebody loses money by investing in an ill-advised derivative instrument, you can be sure that someone else is making money. (We'll put counterparty risk to one side, for the time being.)

In other words, if CDOs are losing money (on a mark-to-market basis) by selling protection, then whoever they sold that protection to will be making money on a mark-to-market basis. And similarly on a cashflow basis: if the CDOs have to pay out when a CDO tranche defaults, they will be paying out to other institutional investors who bought that protection.

So when people like Paul Krugman say that "estimates of the likely losses on CDOs range from $125 billion to $250 billion," they're talking about the amount that the net value of asset-backed securities is likely to fall. They're not talking about the drop in net value of CDSs, because the net value of a CDS – like any derivative – is always zero.

Now if the CDOs start running into serious amounts of trouble, then at that point there could be very nasty problems with counterparty risk. But in order for that to happen, these debt funds would have to have to lose more than all their money. And no one that I know of is yet talking about CDO losses in excess of 100%.

Oh, and one other thing: insofar as CDOs are investing in CDSs and not in actual mortgage-backed bonds, that makes their holdings more liquid, not less liquid. There's a reason why the ABX.HE indices are based on baskets of credit default swaps, rather than baskets of bonds. If it's illiquidity you're worried about, you might not like the market in CDSs, but it's a darn sight better than the market in the underlying MBS tranches.