TRADING DERIVATIVES….Megan McArdle on the banking crisis:
One of the smarter ideas I’ve heard for trying to prevent this sort of thing next time around is putting derivatives on exchanges. Most derivatives are traded over the counter, in part because US bankruptcy law encourages it: as I understand it, derivative counterparties don’t have to get in line with the others, but can seize any collateral they can get their hands on.
Exchange trading enhances transparency, by making it clear what’s out there and roughly who owns it. It also moves the clearing risk to the exchange; while exchanges do fail, they do so much less often than financial firms, and if intervention is needed, they provide a centralized locus for any private or public action.
I’ve been noodling over a list of regulatory changes that ought to be on the table for the next administration, and this is one of them. Credit default swaps, in particular, should be registered like any other security and traded on public exchanges. This wouldn’t make them 100% safe (stock markets have bubbles and busts too, after all), but it would certainly make them a lot safer.
On the other hand, I’d like to hear an argument for even allowing CDOs to exist in the first place, whether they’re publicly traded or not. It’s one thing to chop up securities into different tranches that appeal to different classes of investors — that’s just marketing — but with very rare exceptions the overall yield of such an instrument should be nearly the same as the yield of the underlying securities themselves. A little less, in fact, since you have to factor in additional administrative costs. But the fundamental idea behind modern CDOs is exactly the opposite: not merely that they’re providing a bit of convenience or regulatory arbitrage, but that if you bundle up a bunch of securities and then chop them up in specific ways, they’ll be magically worth much more than the underlying securities themselves. Much more. But this violates a basic law of economics. We’d prosecute for fraud anyone selling a perpetual motion machine, and I’m not sure why CDOs are really any different. Done properly, the market for CDOs ought to be small and sleepy, barely worth anyone’s attention. If it gets non-sleepy, that means it’s becoming fraudulent. So maybe it’s just not worth having at all?
Needless to say, I have no idea how you’d go about banning a particular class of security. And I suppose it’s possible that the underlying problem is with the rating agencies, not the CDO packagers — though real life being what it is, I suspect that’s a distinction without a difference.
In any case, I’d like to hear the argument. Why should we allow the sale and marketing of CDOs at all?