First, a reference portfolio is constructed
Second, Paulson buys protection on the 45%/100% tranche (super senior) from Goldman Sachs, this is a bilateral trade with nothing to do with the Abacus SPV other than using the same reference portfolio.
This leaves Goldman Sachs short protection on the 45%/100% tranche.
Third, Goldman Sachs sells notes to IKB ($150 milion) and ACA ($42 million) => this leaves Goldman Sachs long protection (via purchase from the Abacus SPV) on the notes notional and short protection on the 45%/100% tranche.
Notes sold are well below what was expected in the “flipbook”.
Fourth, post-deal closing Goldman Sachs sells its long protection on the notes (acquired from the Abacus SPV) to Paulson => this puts back Goldman Sachs as short protection on the 45%/100% tranche
Fifth, more than a month after the deal closing, Goldman Sachs buys protection from ACA (through ABN/AMRO) on the 50%/100% tranche, this a bilateral trade with nothing to do with the Abacus SPV other than using the same reference portfolio => this leaves Goldman Sachs short protection on the 45%/50% tranche (5% of total notional).
The ACA deal (50% to 100% = 50%) is for $909 million notional, which implies a total notional of $1.8 billion for the deal, also below the level announced in the “flipbook”.
The deal goes sour and Goldman Sachs exposure is wiped out => $1.8 billion * 0.05 = $90 million
This a synthetic CDO referencing a static portfolio and protection was bought/sold NOT on the entire portfolio notional but only on the super senior tranche and the notes sold to qualified investors.
It is completely irrelevant whether Paulson was or wasn’t an equity investor as the deal doesn’t need an equity investor (and doesn’t have one).
Update:
The final prostectus has now been made public, there is no equity tranche as explained above however my point 5 needs to be amended => the trade is not a bilateral trade but is a partial vertical slice of the unfunded super senior tranche transiting through the Abacus SPV.
Update 2:
Darrell Duffie has produced a nice presentation, the picture below should make everything clearer. (H.T. Morgan_03)
The details of the complexities of the deal are irrelevant to my understanding. This is a disclosure issue. Unless the law explicitly says that qualified investors must be informed if such an investment was put together with the input of the other side of the trade, how could any reasonable person expect that a qualified investor would weigh such information in a material way?
Obviously, the other side of the trade is betting against the qualified investor; whether it is Paulson’s dream portfolio or not seems immaterial. To think otherwise would undermine the purpose to distinguish a person or entity as a qualified investor. (How sophisticated is an investor who relies on such information?) Further, if we are going to say that is it material information, then it follows that we also need to know his reasoning, the extent to which he has offsetting position(s), etc.
In the end, qualified investors need to rely on their ability to analyze the elements of the reference portfolio, not how such portfolio came to be.
(1)-What is meant by a Synthetic CDO short position?
(2)-How did IKB lose that one Billion Dollars? Did it sell CDS protection to GS?
Oh I see (I think) – is it because they employed the “Flipbook” marketing campaign?
Totally surreal.
Thanks jck –
But whether it was Goldman’s first deal with ACA or not – both of those entities were involved in CDOs that were listed on the Irish Stock Exchange, dating from well before this deal and continuing on after it.
Of the 26 Magnetar CDOs identified in ProPublica’s recent piece, the 14 prospectuses all indicate application to list on the Irish Stock Exchange.
So my point is – this deal appears to be an unexplained anomaly – which is very interesting given that it’s the one deal the SEC decided to go after (thus far?). If in fact the prospectus provided is a legitimate document, why was this particular CDO NOT listed on the ISE?
@VenusVictrix:
It was a different deal: first one with ACA out of 22 or 23 under the Abacus brand name. I have no idea how many were listed on the ISE, but of the top of my head I would guess very few (I could be wrong).
Any idea why the prospectus for Abacus is completely different than the format found in all of Goldman’s other CDOs that were listed on the Irish Stock Exchange?
For example, GS’s legal counsel was always Orrick, Herrington & Sutcliffe LLP; AIG-FP appears in the others as Hedge Counterparty, providing interest rate, cashflow, and/or currency swap agreements; they were all listed on the Irish Stock Exchange – among others.
How come Abacus was not listed on any exchange?
Just curious -
@Robert Eberenz:
it’s in the prospectus
My apologies, that was very poorly written…
My question is simple, “what was the ‘reference portfolio’ for abacus?”
My question is, “what was the pool of assets which the ‘imaginary CDS’ contracts were supposed to cover, which ABACUS was then priced to mimic tranches A-1 and A-2 of?” (if that makes any since).
Basically, did goldman just take A-1 and A-2 level synthetic CDO yields from other offerings and then apply them to this deal, or was the price of the ABACUS deal derived upon based upon bid and ask values from IBK/ACA and Paulson? If Paulson and IBK/ACA just came to a comfortable price for an imaginary contract on AAA rated paper, then one would suppose that the value of the supposed CDS’s could be derived…
Your thoughts?
Dolfif, Paulson wasn’t even a particularly well known HF player and what he was known for was merger arb not RMBS structured finance. My bet is that ACA are the ones that thought they were going to take Paulson to the cleaners.
If we believe mr. Pellegrini, senior manager at Paulson, ACA was told that Paulson was taking a short bet on the selected group of names.
The fact that ACA knew to me actually make a lot of sense because it’s almost impossible that the ACA analysts spent days working with Paulson on the composition of the reference portfolio without realizing they were the short better. ACA’s and Paulson’s objectives, clearly stated or not, were opposite. ACA (and IKB) wanted the best possible group of mortgages for their return target. In fact, for the long investor, the return target represent what they will make if everything goes well and the quality of the portfolio represents the risk they are taking for that potential return. Paulsons instead wanted the worse possible group of mortgages for its target cost. In fact, for the short investor, the target cost representes what he is willing to pay for the bet against that pool of mortgages and the quality of the group represents his chaces of being right on the default of many of the names selected.
Many time they must have fund themselves in a the situation where they had two different mortgages, with similar return but one in better shape that the other. Obviously ACA wanted to bet on the survival of the best one, given that for the same return they were adding less risk to the portfolio. Paulson on the other hand wanted to bet on the default of the worse one, given that for the same cost he was getting some-one with higher chaces of default. How did Paulson supported his preference for the mortgage in worse shape? How did they manage to persuade the ACA experienced managers to accept the worse name for the same return of a better mortgage??
They must have know Paulson intention and not believing that the market was going to collapse (like many still did at the beginning of 2007) they thought that the combination of the return offered by the deal and group of names agreed at the end was good enough to take the other side of the bet.
Let’s not forget that Paulson was yes a well known HF at the time, but it wasn’t considered “one of the smartest around” a status it only acquired AFTER getting the collapse of the mortgage market right.
@mister_x:
ACA would have been easily disabused by reading the flipbook and the prospectus, the former shows “first loss” 0%/10% tranche, not offered, no fee (for ACA, that matters since they received fees on the other tranches except the SS), and no coupon, the latter which describes the final structure shows no equity tranche, there is a small first loss tranche, unsold but it’s related to something else => the collateral securities and I want to keep it simple so won’t get into that because it’s irrelevant.
For a moron like me I need an explanation but have read the complaint and listened to the Goldman quarterly conference call where the general counsel offered a defense. I think these facts are agreed upon.
At the end of the day Goldman marketed securties witout mentioning this portfolio to long investors without disclosing the securities had been offered up by someone who had paid Goldman to distribute his dream short.
Goldman says the above statement did not need to be disclosed because of the presence of ACA. If the presence of Paulsen was beign why did Goldman refuse to disclose to ACA or more importantly in the offering material? Their defenders say because of the presence of ACA.
But what if 12 jurors decide that trying to sell a portfolio that was constructed with the intent to go down to a long investor is a material fact worth disclosing? After O.J. I stopped trying to predict juries but especially in a civil case where the standards of proof are lower I think Goldman is in trouble.
“ACA was never explicitly disabused of this notion once they perpetuated it”
*By “they” I mean Tourre/GS of course.
jck, this may be the SEC’s strongest point (in what has been disclosed so far):
“Fabrice Tourre‟s January 10, 2007 e-mail to Ms. Schwartz containing the “Transaction
Summary” in which he stated that the transaction was “sponsored by Paulson” and
included the line: “[0] – [9]%: pre-committed first loss,” (GS MBS E-003504901) which
the Staff stated described the equity tranche”
Even though the final prospectus doesn’t mention the equity tranche, the pitchbook certainly supports the above statement. ACA was never explicitly disabused of this notion once they perpetuated it, which is the SEC’s contention and Felix’s. Even though the presence/absense of the equity tranche was immaterial to the actual performance of the CDO, isn’t this misleading?
Even GS knows this could be big trouble and their arguments namely, (a) Tourre doesn’t remember (b) Try to obfuscate the meaning of “sponsor” (c) Claim Kreitman knew nothing (!), are EXTREMELY weak and could be their undoing yet.
Disclosure: No horse in the race. Just interested in the case.
@Sandrew:
The SEC basically alleges that because ACA had done previous CDOs were the shorts super senior also bought the equity tranche, Goldman Sachs should have told them that their assumption was wrong this time. Sorry, that’s ridiculous, they don’t have to do that, the flipbook and prospectus are clear => no equity tranche.
As for the information regarding Paulson’s position, plenty of reasonable people will argue against the SEC that it’s not material for a very simple reason => whoever is long or short has no bearing on the performance of the notes which depends only on macro economic factors and whether people pay their mortgages or not. And no investor is mandated to disclose his overall position in that kind of deal. If I buy XYZ stock from you, do I have to tell you if I am shorting calls on XYZ or not => no. Same here.
Out for the day, this conversation will have to be temporarily interrupted. Nice talking to you.
“Whether Paulson is long equity outside the Abacus is irrelevant” is a statement about which reasonable people disagree. SEC alleges that Paulson’s net position in the Abacus reference portfolio (irrespective of their view on ABS-backed CDOs generally), in combination with the fact that Paulson had input into the composition thereof, was material information to ACA and IKB.
For the record: I find the SEC’s case pretty weak. I’ve yet to see a smoking gun showing that Goldman conspired to mislead ACA as to Paulson’s net position. In the meetings between ACA and Paulson, ACA ought to have been introduced to Paulson as the protection buyer, which is exactly what Pauslon alleges. ACA should not have been spooked by this. Clearly both the longs and shorts in a bespoke basket must come to common agreement on both the portfolio and the pricing.
@Sandrew:
You are moving the goalposts.
Here is what Felix wrote:
“Here, then, is arguably Goldman’s biggest lie of omission: it never told ACA that the equity tranche didn’t exist.”
It didn’t exist within the Abacus deal. End of story.
Whether Paulson is long equity outside the Abacus is irrelevant, he can do what he wants, he could have bought/sold some ABX tranches and that’s none of the Abacus investors business or Goldman for that matter. Neither Paulson or Goldman have to disclose their overall positions, they can hedge or not, that’s their business, not the business of the Abacus investors.
My only observation is that the disclosed absence of investors in a funded equity tranche does not rule out the possibility that Paulson could have been synthetically long equity. Felix/SEC view not dead yet.
@Sandrew:
Felix specifically mentions the equity tranche not “a lower tranche” and neither the flipbook or the prospectus support the view that the equity tranche was for sale.
I’m not so sure about destroying Felix’s theory, and to which I don’t yet subscribe. True, it is clearly shown that there was no investor in the equity or other lower tranches. But ACA was well aware that others were (or could have been) taking synthetic long positions in the structure on the side. (This is Felix’s “ontological status” argument.) If ACA believed that Paulson was planning from the get-go to be long a lower tranche, nothing in the prospectus would have necessarily disabused them of this notion.
@David Harper
Thx, a picture is worth a thousand words ;-)
I still am not a fan of using “unfunded” for the unsold tranches, => the super senior is unfunded but somebody wins or loses depending on performance while the other classes are unsold i.e. nobody bought them and nobody will win or lose from their performance.
Also the protection buyer facing the SPV is Goldman Sachs, Paulson’s counterparty is Goldman not Abacus (this is going to matter when going to court)
I read the prospectus quickly and it does destroy completely the Felix Salmon big lie theory => there is no equity investor and ACA would have known that by simply reading the prospectus before signing on the dotted line.
I illustrated the final, stylized structure @
http://www.bionicturtle.com/learn/article/goldmans_abacus_structure_illustrated/
sorry, jck. could you put together another post for complete morons w/out acronyms and assumptions that readers know wtf you’re talking about? I can’t pass this on to my mom.
@jck: Thank you for working so hard to demystify this stuff for the masses!
@csissoko:
He paid about (or slightly more than) 150bps on his share of the notional (i.e. arranging the $909 million super senior), not on the total notional of Abacus, which is $1.8 billion. The fees on the CDO tranches were paid to ACA by the protection buyer interfacing with the Abacus SPV i.e. GS, these were very small since little of the tranches was sold except to IKB and …ACA.
Just a (possibly stupid) question: If the Paulson CDS wasn’t part of Abacus, why did Paulson pay Goldman a fee based on the total notional of Abacus (I think I remember you telling Felix Salmon that that’s how it would have been calculated)?
@Sandrew:
1. Right about IKB, it’s $150 million, I corrected the post ( a moment of inattention…)
IKB was in the triple A and ACA in the double A, these 2 tranches were likely vertically sliced, i.e attachment points roughly similar to the first prospectus but they were not fully subscribed. This doesn’t matter because GS was only counterparty to what was actually sold (via the SPV).
2. We don’t know the exact timing of the bilateral trades GS/Paulson and GS/ACA/ABN but that not really relevant because these are bilateral trades not related (not going through) the Abacus SPV and not pertinent to the offering circular, I suspect the GS/Paulson was done way before the GS/ACA/ABN closed but that doesn’t mean there was a gap for GS, the latter trade was complicated by the credit issue (ABN having to back ACA) but the commitment may have been made before, however I don’t know that for sure.
3. In synthetics, anything is possible, the reference portfolio is just that, you can structure against all or part of the total notional, it is common not to cover the full notional in smaller synthetic but it is not widely known because these are usually “bespoke” tranches sold via bilateral trades that don’t require an offering circular and are private.
There is absolutely no reason to cover the full notional to do a synthetic, in the Abacus case the equity tranche was never offered, and obviously what was offered was way undersubscribed.
Possibly the clearest explanation I’ve seen anywhere. Thanks.
Thanks, jck. I’ve been toying with this timeline since Monday, and came to largely similar conclusions at my last comment here: http://bit.ly/c59bhI
Couple of nits:
1. I believe IKB purchased only $150m of notes, which together with ACA’s $42m totals the $192m. I also recall reading something from one or the other of Goldman’s defense letters that suggested that these were different classes of notes (though I’m not sure if that necessitates that they have different attachment/detachment points).
2. Do we know the timing of the 45-100 bilateral trade between Goldman and Paulson? If so, from what source? Would Goldman’s famed risk managers have permitted such an exposure prior to putting on a hedge (ultimately transacted, in part, in the ACA/ABN transaction)? (Note: I don’t mean this is to be a rhetorical question, it may indeed be so even if it is a bit stupid.)
3. While I recognize that there’s nothing to structurally mandate that total portfolio notional be consistent between the funded notes and each of the two bilateral supersenior trades, I had been working under this presumption (in the absence of evidence to the contrary). Is that common?
Sandrew
finally, some people who actually understand things.. thanks guys!
Yep, maybe, but could also be that the notes were sold as vertical slices i.e. the triple A was expected to be $480 million (21%/45%) and $150 million was sold to IKB, maintaining the same attachment parameters. In other words the attach/detach points need not be moved because the entire tranche wasn’t sold => what matters here is that protection sold equals protection bought, regardless of what was sold. The CDSs were not written on the reference portfolio, but on pieces of the reference portfolio.
All true – the only addition I would make is that the att/det points on the notes (whatever they were to synthesize the $150m IKB and 42m ACA exposure to the pool) would have been equally geared to the implied $1.8bn notional – ie 150m at risk for IKB = 36.67-45 att/det, or something like that. Basically solve for attachment points to whatever the clearing yield was to arb the AAA rating out of the agency.