When Papers Promise But Don't Deliver, CDO Edition

I continue my search for someone who can shed light on exactly what happens to the alphabet soup of MBS and CDO and CDS when subprime default rates rise. I briefly thought I'd hit paydirt when Alea blogged a February paper by Joseph Mason and Josh Rosner, with a very enticing abstract:

The authors go on to measure the efficacy of ratings agencies when it comes to assessing market risk rather than credit risk. They determine that even investment grade rated CDOs will experience significant losses if home prices depreciate...

Was this what I've been looking for, an explanation of the link between home prices, mortgage-backed bonds, and CDOs? Alas, it was not to be. I've scoured the whole paper two or three times now, and – well, I just can't find their finding. Home prices are barely mentioned in the report, and not at all in the part of the report which deals with CDOs. I also can't find any mention of losses in investment-grade CDOs for any reason at all. So I certainly can't find any link between home prices and losses in investment-grade CDOs. The closest thing I can find is this:

Given recent events, we now know the defaults are in the mortgage pools and it is only a matter of time before they accumulate to levels that will threaten rated mezzanine RMBS. Given the high proportion of CDO investments in mezzanine RMBS, the questions therefore become: (1) when will the defaults hit CDO returns and (2) what will be the effect when CDO investors react, as they did with previous sectoral difficulties, by divesting the sector and moving on to new forms of collateral?

The statement that "it is only a matter of time" before losses hit the MBS market is especially weird in light of the fact that they haven't mentioned this before, and that they say earlier on in the paper that as far as the MBS market is concerned, "prepayment risk is an entirely different order of magnitude than default risk". It's as though they can't make up their mind about default risk, and whether it's a tiny problem or a massive one.

But in any case, all of the authors' discussion about risks in the CDO market centers on the probability that CDOs will move out of subprime mortgages and into some other asset class, reducing liquidity in the MBS market and ultimately hurting the housing market as a whole. That's all perfectly reasonable – but it doesn't even come close to demonstrating that existing CDOs with investments in the MBS market will suffer significant losses on their rated tranches.

So, the search continues. In the meantime, there do certainly seem to be winners as well as losers from the current mess: a $2 billion fund run by John Paulson returned 40% in June alone – and that's after fees. I guess that losing his fight against Bear Stearns didn't do him any harm at all.

And I hasten to add that I've nothing against commenter jck, who runs the Alea blog and who left a very handy comment on my last subprime post. According to him, the implied spread on BBB- rated bonds, if we can trust the ABX index, is about 2500 basis points. That's huge: debt securities very rarely trade at that sort of level without defaulting. On the other hand, securities at the bottom of an asset-backed waterfall do exhibit a huge amount of price volatility, since a small change in the structure's total cashflows can mean a huge variation in recovery value.

Update: jck emails me to point out that a later version of the paper has more detail – although it still doesn't draw any connection between house-price declines and CDO defaults. He adds:

I think they are misguided if they think downgrading the asset side automatically implies downgrade of the liability side of the CDO, for example if there is a general credit degradation with increased correlation, the junior tranches will actually gain against the senior tranches i.e spreads will contract, but this a draft so I will wait for the final print before getting more critical.