In the various blogosphere efforts to dissect the Goldman Abacus transaction now in the SEC’s crosshairs, some commentators have characterized it as unusual, a “bespoke” CDO“, or “a very complicated deal…a supersynthetic CDO.” Effectively, the view is that Abacus is a multi-tranche variant on the single tranche CDO structure that was developed with corporate CDS, well before CDS on asset backed securities existed.
I was puzzled by these descriptions, because based on my reading of the prospectus, the notorious flipbook, and the SEC’s claim, I didn’t see anything terribly unusual, save that ACA was the manager and guarantor and some of the long exposures never were placed, meaning Goldman wound up stuck with them. Moreover, it seems likely that the Goldman suit is a test case for the SEC, and if it succeeds here, it may well launch other actions. Hence it would not serve the SEC to target an unusual deal. But I have thought a lot about hybrid CDOs (where some tranches are funded, some partially funded), not much about synthetics, so I could easily be wrong. I wanted to getting a reading from someone who actually puts these deals together, and an expert obligingly weighted in:
You’re right, any CDO could rightfully be called a “bespoke”… the underlying portfolios are all different. But the nomenclature “bespoke” implies something quite different when it is used in the structured product community. “Bespoke” deals are much more complicated and the term is much more often used in the CSO market. The typical “bespoke” deal is an n-th to default basket. An investor comes to a correlation desk and says: “in a basket of 100 IG corporate CDS, I want to take exposure on the 9%-15% slice.” The underwriter/arranger will take and (try to) hedge the risk on the remainder of the basket (i.e., 0-8 and 16-100).
This was not ABACUS. ABACUS was a normal CDO… just in unfunded form.
I think people are confused by the fact ABN’s name has been thrown into the mix. For their most part, their involvement is irrelevant to the CDO itself. One of ACA’s business lines was bond insurance. In this particular CDO, instead of funding the super senior tranche (i.e., creating a note and placing the proceeds from that sale in escrow, earning LIBOR, held there in case losses on the underlying portfolio required a payment to the protection buyer on the underlying portfolio of CDS), ACA essentially wrote protection on it to Goldman. Its the same flow of moving parts as a normal monoline policy… its just the note itself on which the policy was written was never sold/funded.
Goldman refused to take the credit risk of ACA so they did a back-to-back swap with ABN… where for a few basis points ABN agreed to take the credit risk of ACA not being able to pay on the policy and became the counterparty facing Goldman. I hope this makes sense… what it does is essentially make ABN the owner of the super senior tranche, with the benefit of a monoline insurance policy written by ACA.
The funded tranches of a deal like ABACUS — the bonds sold to IKB, the bonds sold into other CDOs, and the bonds Goldman was stuck with — act like the upfront margin in a standard CDS contract. Think of it this way…
I’m a hedge fund and I write protection on a Baa3 rated subprime bond currently trading at $100-00. The price implies the market’s current expectation that the bond will take $0 of loss… BUT, I’m a hedge fund selling protection on something rated Baa3 so my counterparty is still going to require me to post some upfront margin/collateral. That same idea applies to the funded notes in a synthetic CDO. The proceeds from the sale of the funded notes is placed in escrow (on a side note, the “escrow” accounts I’m referring to blew up as well… AIG/FSA did this business and took the money and invested it in subprime securities) and used as collateral against potential losses on the underlying CDS portfolio.
Nevertheless, ABACUS was a “plain vanilla” synthetic CDO and the ONLY reason Goldman was stuck with any long exposure was because it was a failed underwriting. They would have placed the bonds if they could but no one would buy them at the price at which they were offered.