In case you haven’t figured it out, Ed’s post on Sweden highlights an important and troubling development. It seems that central banks have locked themselves into “if the only tool you have is a hammer, every problem looks like a nail” behavior with the move to negative rates. Since they are best able to dispense liquidity, well, it is liquidity you will get. The assumption, as has become conventional wisdom among the economic elite, is that the reason Japan stayed mired in borderline deflation is that it wasn’t aggressive enough with quantitative easing and fiscal stimulus. The Japanese have a rather different view, since their economy actually did show good growth in 1996, but then they got into “it’s all over” mode and raised taxes. The Asian crisis didn’t exactly help either. Their assessment, which they have oft repeated, is their big mistake was not cleaning up the banking system sooner. The Japan mess had a different dynamic (it made no sense for banks to lend because there was so much money on offer relative to willingness to borrow, at least among banks that were worthwhile credits that spreads on lending were close to nil and banks could not recapitalize on such thin spreads. Not exactly a problem we have here).
Now to my pet issue. I feel as if I am in blind man and elephant mode, and that we observers of the crisis are collectively in that fix. Which raises an interesting philosophical question: if blind men who encountered an elephant realized the strange critter was bigger than any of them could dimension individually, could they have come up with an adequate description of an elephant among themselves?
The immediate question was triggered by the Michael Lewis piece on AIG, in which he pointed out that AIG quit writing CDS on subprime in very early 2006. The most superheated phase of subprime lending was third quarter 2005 to end of 2006. Although crappy subprimes were being sold in 2007, the volume fell off as the product got a bad name.
Sales of product involving subprimes involved credit enhancement. So if AIG, the biggest stuffee, dropped out, what replaced them? Lewis suggests “Wall Street took the risk”. It’s a throwaway comment, and I don’t think it’s accurate, or at best only partially accurate.
Here are some obvious places the risk could have been absorbed. If readers can provide any feedback (as in do they know of a marked increase in any of these approaches in 2006 v. 2005), it would be very useful. I have another idea I’ll volunteer later, but am curious as to any datapoints re these possibilities:
1. Originator takes more risk, via agreeing to cure a higher level of defaults in the pool. In theory, this can take place via substitution, but in practice, it means the originator takes a loss (swapping good paper for bad). Thus this means the credit enhancement ultimately depends on originator credit quality. This seems a pretty weak form of enhancement, and in theory worsens the appeal of the deal to the originator, but if everyone was punch drunk on risk, maybe no one noticed such niceties.
2. Other insurers step in, like MBIA and Ambac
3. Foreign stuffees. Landesbanken bought a ton of this paper, but I am not certain how much it would have done re the credit enhancement (getting the needed ratings)
4. Overcollateralization. That effectively means making the equity and mezz tranches bigger, in terms of total value, relative to the rest of the deal. Any evidence that that happened?
And a separate question: who was buying the equity layer of CDOs? I have read the investment banks retained it (the sense I had was this was sort of free money, they made enough whether the equity bit paid off to be indifferent). However, i have also seen it argued that hedge funds were big buyers, the theory being that the CDOs overpriced the AAA layers, hence the equity was cheap. Any sense (or better yet, estimates) of how much CDO equity went to hedge funds? This stuff had amazingly high embedded leverage.
One reader did say that the CDO structures shifted to leveraged super senior CDOs (remember CDOs took the mezz portion of subprime deals, which means they are signing up to be hit early on with the effects of defaults. Having some one else take first or early losses enhances credit for everyone else) and they kept the super senior layer. I hate to be a bit dense, but I don’t see how taking the LEAST risky bit helps in risk absorption.
And my impression was keeping the super senior layer was a bit of an accident, as in they couldn’t sell enough, but figured what the hell, this is decent paper, we can keep it (they could repo it for no haircut until everybody got nervous re subprime). That’s the sense the media gave re Merrill, which retained a lot of super senior CDO paper.
OK, now to my pet theory as to where a lot of the risk went: synthetic CDOs. Remember, synthetic CDOs are composed from the cash flow of CDS written on other risks. Presumably a lot of those CDS were subprime deals, either the original MBS or the CDOs they went into (again, remember CDOs generally contained a lot more than subprime RMBS, they were a big home for collatearlized loan obligation tranches, but the subprime deals were particularly helpful in making those structures work).
So what exactly is “the cash flow from CDS” mean? Um, you are the insurer. You are taking the periodic CDS payments and will have to pony up if the credit goes bad (or with a higher tranche, bad enough so as to breach a threshold level).
Synthetic CDO issuance was nearly as large in aggregate as cash CDOs.
Does that add up? One clue would be if synthetic CDO issuance picked up considerably in 2006 v. 2005, that would suggest it was primarily a way to lay off risk (as opposed to the professed reason, “there was so much demand for this paper, we had to create synthetics to meet demand”. I am of the “stocks are sold not bought” school as far as funky paper is concerned).
And if so, who in their right mind was signing up for this stuff? Particularly the equity and mezz tranches, which is where the serious risk assumption was taking place.
Thanks! If you aren’t set up to comment but have ideas or info, please ping me at firstname.lastname@example.org