Felix Salmon

Goldman’s unwanted super-senior position

By Felix Salmon
April 21, 2010

Kate Kelly has the obvious answer to the question of what on earth Goldman Sachs was doing with that 45-50% super-senior tranche:

“The core of the SEC’s case is the allegation that one employee misled two professional investors by failing to disclose the role of another market participant in a transaction,” said Goldman CEO Lloyd Blankfein said in a recent message to employees. The firm “assumed risk in the deal, and we lost money,” he added.

But Goldman invested the money only because sales of the deal didn’t play out as planned, forcing Goldman to step up with its own money, people familiar with the matter say.

This is surely exactly right. We’ve already seen that it took Goldman a full five weeks after the deal closed just to wrap the 45% of the super-senior tranche that it did manage to get off its books — and it did so at the end of May 2007, when the mortgage market was hitting three-month highs and there was a lot of hope in the market that the early-2007 crash in the ABX index had been a short-lived aberration.

Here’s my theory of what happened: Goldman intended to wrap the super-senior tranche with ACA more or less at exactly the same time as the deal closed. But it couldn’t come to terms with ACA on the question of posting collateral: ACA, which only had a single-A rating even then, wanted a Berkshire Hathaway-style deal whereby it didn’t need to post collateral on a day-to-day basis and just needed to pay out once there was a credit event. That wasn’t acceptable to Goldman’s risk managers, which spent a good amount of time putting a deal together whereby ABN Amro would buy the wrap from ACA, and in turn would sell insurance to Goldman while posting collateral as necessary.

But the price rose when Goldman had to go through ABN Amro, and so Goldman ended up offloading only the top 45% of the 50% super-senior tranche. The rest it kept on its own books as a trading position, looking to sell that unwanted long position on an opportunistic basis.

And then Goldman ran out of time: before it could sell the position, the structure had started imploding, and Goldman was stuck with $90+ million of losses.

Which now Blankfein and Goldman are trying to turn into some kind of virtue.

This is one of those things a bit like the argument saying that ACA rejected half of Paulson’s picks: making the argument involves making a subtle implication that the case would be stronger were these facts not true. Is Goldman saying that the SEC’s case would be stronger if ACA had not rejected any of Paulson’s picks? Is it saying that the SEC’s case would be stronger if Goldman had managed to wrap the whole super-senior tranche and had ended up making a profit on the deal? Because it’s pretty obvious that Goldman would have been perfectly happy with the outcome in both those cases, and still wouldn’t have disclosed anything more than it did.

Of course, you can’t sue a bank for not disclosing information in a hypothetical deal which never actually happened. But the legal case is only half the point, and not even the more important half. The ethical case is what matters, and Goldman’s doing a very bad job of answering that one.

9 comments so far | RSS Comments RSS

so you think they actually lost money on this trade, then? and not that they’re mincing words, and saying they lost this $90mm, but actually made that up by hedging the tranche with CDS that’s technically not part of this CDO transaction?

Posted by KidDynamite | Report as abusive

Apparently Fabulous Fab earned a 2 million at the end of the year. Did Goldman him reward him with a giant bonus for losing the firm 90 million? Unlikely! That statement about losing 90 million looks like another giant present lie!

Posted by DanHess | Report as abusive

Blankfein should be more careful with statements like that. That rah-rah stuff may rally the troops now, but if some of it turns out to be not entirely true – like they “assumed” risk in the deal, rather than got stuck with an unsold stub – and the team will turn on him. This is a high stakes game, internally and externally. I’ve seen first hand this sort of thing blow up on CEOs before – shading the truth to pump people up, create an us vs. them environment. It’s ugly when the real truth comes out from an outside source.

Posted by tja3 | Report as abusive

What was that odd bit about ABN AMRO paying 27mil to GS for insuring AcA’s credit risk, if I recall correctly? I never understood that part.

Posted by mister_x | Report as abusive

@KidDynamite, 90mm seems to be just the headline number, what they lost on the tranche. As you say, even if the desk didn’t hedge the position, the firm surely had some sort of hedge on. But the biggest mitigant would have been the large first-profit position they had as deal arranger.

Posted by Greycap | Report as abusive

Do we know when Paulson’s short of the 45-100 tranche occurred? Note that this piece of Paulson’s trade doesn’t appear to have been part of the ABACUS structure per se (though it is surely related, as they’re both connected to Paulson’s reverse inquiry to purchase protection on senior tranches of a subprime mezzanine backed CDO). That is, as far as I can tell, Goldman never purchased protection on the 45-100 tranche from ABACUS 2007-AC1, since I imagine no one was willing to fund this large, low-yielding tranche. I was working from the presumption that the ACA/ABN 50-100 trade occurred concurrently with the Paulson 45-100 trade. If the timing was off, this indeed would go a long way to helping explain how Goldman retained the strange 45-50 semi-supersenior tranchelet.

Posted by Sandrew | Report as abusive

Sandrew, Paulson didn’t buy protection on Abacus. He bought protection on a bunch of Reference Securities, which would pay out $1 billion or so if they all went to zero. The first loss there, the first payments of 0-$90 million, were equity, then the next ones were single-A, etc. The super-senior exposure always existed, the only question was who ultimately had it.


mr salmon — you’re incorrect. the structure of the deal is as follows. Goldman is Abacus’ counterparty on the CDS for Abacus’ underlying portfolio. Abacuse provides Goldman with protection on the underlying reference entities, and Goldman pays Abacus for that. Paulson comes in once the CDO has been created, and obtains protection on the resulting CDO tranches. why all this elaborate effort, when he could have just obtained CDS on the reference entities directly? because it was far more expensive to do that — the reference entities were BBB-rated. instead, the CDO’s value entirely depends on those BBB-rated underlying tranches, but the CDO tranches get higher ratings. so Paulson has to pay less for a CDS on a CDO AAA tranche (even though that tranche depends for its value on all those BBB tranches) than he would have to pay for CDS on the RMBS BBB tranches themselves. whatever else it may have been, it was really, really smart.

Posted by oriduck | Report as abusive

What’s your source on that, Felix? I’ve been trying to work out the who-held-what-when questions since Monday, and I’ve yet to see anything compelling to suggest that Paulson managed to short (i.e. buy protection on) the *entire* capital structure (0-100) of the ABACUS 2007-AC1 reference portfolio.

Paulson probably would have loved to have purchased protection on the whole thing, but Goldman wouldn’t likely sell it to them unless it had someone on the other side on whom to offload the risk. From what I can glean from reports/allegations/defense docs/bloomberg, the ABACUS deal only had $192M of securities issued (to only IKB for $150M and ACA for $42M). I believe these securities represent what is effectively a 34.4%-45% tranche on a $1.818B reference portfolio. Incremental to this (but never passing-through the ABACUS legal entity) there was a supersenior (50-100) trade between GS and ACA/ABN to the tune of $909B. (Note: this is where I get $1.818B = $909Bx2). Of course, Goldman sold all of that protection (and a little bit more) to Paulson.

If you’re wondering where all the lower tranches (the 0-34% first-loss risks) went, stop wondering. If my suspicious are correct, it never existed. The lower tranches referred to in the pitchbook never found a way to market because Goldman couldn’t find any investors. Now, mechanically, there are a number of ways to accomplish this feat; one way is for Goldman to enter into a single-tranche bespoke basket default swap facing the ABACUS SPV (in the first instance, but this was ultimately backed-to-backed with Paulson). A single-tranche bespoke basket default swap is basically a customized tranche (e.g. 34.4% to 45%) of an unfunded synthetic CDO all wrapped up in an ISDA-friendly derivative contract. It’s a bilateral derivative agreement with a specified reference portfolio (usually it just points to the deal docs for the notes) and attachment/detachment points.

Posted by Sandrew | Report as abusive

Post Your Comment

We welcome comments that advance the story through relevant opinion, anecdotes, links and data. If you see a comment that you believe is irrelevant or inappropriate, you can flag it to our editors by using the report abuse links. Views expressed in the comments do not represent those of Reuters. For more information on our comment policy, see http://blogs.reuters.com/fulldisclosure/2010/09/27/toward-a-more-thoughtful-conversation-on-stories/