Original posted on Reuters by Felix Salmon:
Alea has given us Abacus for Dummies: a very useful quick overview of the structure of the deal. And in doing so, he helps to reveal Goldman’s biggest lie.
Simplifying Alea even further, we have five steps here:
- The reference portfolio is put together.
- Goldman sells super-senior protection, Paulson buys it.
- IKB and ACA sell senior protection, Goldman buys it.
- Goldman takes the senior protection that it bought from IKB and ACA, and sells it on to Paulson.
- Goldman buys super-senior protection from ACA, through ABN Amro.
From Goldman’s point of view, steps 3 and 4 cancel each other out as a perfect hedge, and it can walk home happily with its fee income. And steps 2 and 5 should do the same thing, but they don’t: in step 2, Goldman sold super-senior protection on the top 50% of the deal, while in step 5 it bought super-senior protection only on the top 45% of the deal. So the hedge was imperfect, Goldman ended up long 5% of the deal, and, in the end, it lost lots of money.
But the fact is that there are two big-picture deals here, not one — and yet they’re very intimately connected.
In the first deal, the super-senior deal, Goldman acts as an intermediary, first selling protection to Paulson, and then buying it from ACA via ABN.
In the second deal, the actual Abacus deal, Goldman creates the Abacus vehicle, which issues securities to IKB and ACA, which are ultimately funded by Paulson taking the other side of the trade.
Now, let’s look at a page from the pitchbook:
The Super Senior tranche here is the one in the first deal; Class A is the notes which were sold to IKB; Classes B, C, and D were sold to ACA; and the First Loss equity tranche at the bottom is the bit of the deal which never existed since, as Alea says, “the deal doesn’t need an equity investor (and doesn’t have one)”. The fact of its nonexistence is conspicuous by its absence: “Not Offered” is by no means the same thing as “Does Not Exist”.
The fact is that the Abacus deal itself didn’t need a super senior investor, either, and didn’t have one. It just needed classes A through D, which were sold to ACA and IKB. The super-senior deal was entirely separate, and had nothing to do with Abacus, although it used the same portfolio of reference securities.
So the question arises: why on earth is the super senior tranche (and the equity tranche, for that matter) even listed in the pitchbook in the first place?
When Goldman refers to ACA as “the overwhelmingly largest investor in the transaction”, it’s clearly referring to the transaction as a whole, including the super-senior deal, rather than just the Abacus part of it. And what’s more, there’s something clean and elegant about the way in which the structure of the deal, as outlined in the pitchbook, goes smoothly all the way from First Loss all the way to 100%.
In theory, as far as I can tell, there’s no reason why the 45-50% tranche — the one that Goldman ended up holding onto and losing $90 million on — should ever have existed either: it, like the equity tranche, could simply never have been offered to anyone. Why didn’t Goldman just move the attachment point for the super senior tranche up from 45% to 50%, so as to match the hedge it bought from ACA via ABN?
After all, Paulson was not buying credit protection on the reference securities as a whole. But that’s how it looks, in the pitchbook. Here’s the pretty picture of the structure, in the same book:
If you’re ACA, looking at this structure, you know that as the deal is being put together, you’re negotiating to insure the Super Senior Amount which exists in this picture as a semi-fictional entity outside the structure and inside a grey dotted box. In other words, while you know it’s not a formal part of the Abacus structure, you also know that it exists.
And in this picture, the First Loss Amount has the same ontological status as the Super Senior Amount: it exists, but only outside the formal confines of the Abacus deal. Since ACA knew full well that the super-senior tranche existed — after all, it was negotiating to insure it — there’s no reason why it should have doubted that the equity tranche existed as well, just like it did in Magnetar trades with which it was familiar.
Here’s Goldman Sachs, in its letter to the SEC:
The fact that ACA may have perceived Paulson to be an equity investor is of no moment. As a threshold matter, the interests of an equity investor would not necessarily be aligned with those of ACA or other noteholders, and holders of equity may also hold other long or short positions that offset or exceed their equity exposure. Indeed, Laura Schwartz of ACA understood this from her work on a transaction that closed in December 2006 in which Magnetar, a hedge fund that bought equity and took short positions in mezzanine-level debt, participated. (See GS MBS-E-007992234 (“Magnetar-like equity investor”).)
This is, I think, inadvertently damning to Goldman’s case. It’s true that in the Magnetar deals the entity which was long the equity tranche was also short the debt. But at the same time, it’s also true that in the Magnetar deals there was no question that the equity tranche existed. So if ACA’s Schwartz was thinking in terms of the Magnetar deal which closed just before the Abacus deal started being negotiated, then it’s quite understandable that she believed in the existence of an equity tranche in this deal, too. And Goldman never did anything to disabuse her of that belief.
The clear message of the pitchbook is that this synthetic CDO was put together to mimic a cash CDO, which has to have all of its tranches spoken and accounted for. You can’t have a cash CDO without an equity tranche. Remember that if the only point of the Abacus deal was to create the Abacus securities, then there wouldn’t have been a super-senior tranche at all, and ACA would not have been the largest investor in the transaction.
Here, then, is arguably Goldman’s biggest lie of omission: it never told ACA that the equity tranche didn’t exist. If it was being a true and honest broker, it should have done. End of story.