The Goldman and Magnetar letters

By Felix Salmon
April 20, 2010
Goldman's first letter to the SEC:

" data-share-img="" data-share="twitter,facebook,linkedin,reddit,google" data-share-count="true">

Goldman’s letters to the SEC make for fascinating reading, and I’ve learned quite a lot from them, including where Goldman’s $90 million loss came from: tt’s not an unfunded equity tranche at all!

Here’s Goldman’s first letter to the SEC:

Because Goldman Sachs purchased protection from ACA on a portion (50-100%) of the super senior tranche, but wrote protection to Paulson on the entire (45-100%) super senior tranche, it bore the risk that poor performance of the Reference Portfolio would affect the 45-50% portion.

It’s a bit weird, that Goldman would leave itself with the 45-50% tranche of the Abacus deal, but that seems to have been the case.

More generally, the Goldman defense — which is as sophisticated and well-argued as you’d expect from bankers and lawyers of this caliber — seems to be that ACA was an enormous asset manager which neither knew nor cared about a small fund manager like John Paulson. If they thought about his positioning at all, they would probably have come to the conclusion that he was short, and if they came to that conclusion it wouldn’t have stopped them going forwards with the deal, because they considered themselves to be highly sophisticated when it came to putting together CDOs.

It’s a reasonably strong argument, but it fails utterly to answer the question of why on earth, in that case, Paulson’s role wasn’t disclosed much more transparently. The whole deal came out of a reverse enquiry from Paulson to Goldman: why couldn’t Goldman, bringing ACA into the loop, explain the whole concept in the space of a couple of minutes? Why all the studied ambiguity about equity tranches and sponsorships? Why not just come out and say that Paulson wasn’t taking an equity slice, and was going to be short the entire structure?

Here, for instance, is the disclosure from Magnetar’s Auriga deal:

Initial Preferred Securityholder may enter into credit derivative transactions relating to Reference Obligations or Cash Collateral Debt Securities in the Issuer’s portfolio. On or after the Closing Date, the Initial Preferred Securityholder may enter into credit derivative transactions relating to Reference Obligations or Cash Collateral Debt Securities in the Issuer’s portfolio, under which it takes a short position (for example, by buying protection under a credit default swap relating to such obligation or security) or otherwise hedges certain of the risks to which the Issuer is exposed. The Issuer and Noteholders will not receive the benefit of these transactions by the Initial Preferred Securityholder and, as a result of these transactions, the interests of the Initial Preferred Securityholder may not be consistent with those of Noteholders.

That kind of thing would have gone a long way in the Abacus disclosures — and remember that in the Auriga deal, Magnetar really was long the equity tranche. Goldman desperately tries to say that this disclosure wasn’t worth including because it doesn’t identify Magnetar by name — but a disclosure like this would not have needed to identify Paulson by name, either, in order for ACA to know exactly who it was talking about.

Meanwhile, John Gapper has obtained Magnetar’s letter defending itself from ProPublica’s allegations, which is well worth reading, and which is plausible on its face:

Magnetar’s strategy was in essence a capital structure arbitrage. This type of strategy is broadly employed in corporate credit markets, and is based on the relative value between differing components of a company’s capital structure (in our case the different tranches or classes of a CDO), and on the supply‐demand imbalances which can be exhibited in the pricing of rated and non‐rated tranches. From early 2006 to late 2007, there was a systematic relative value mispricing between the equity tranches of Mortgage CDO structures, which offered approximately 20% target yields, and mezzanine debt tranches of Mortgage CDO structures, on which credit protection could be bought for between 1% and 4% (depending upon which tranche and CDO).

If anything, the story of the Abacus deal makes the Magnetar letter more believable, since if Magnetar really just wanted to go short subprime securities, it could have done so Paulson-style, without taking on the equity tranche at all. ProPublica never really demonstrates that Magnetar was in the business of making Paulson-style directional macro bets, as opposed to clever relative-value plays which paid off very well when correlations unexpectedly went to 1.

That doesn’t mean that the Magnetar Trade was all sweetness and light, however. In trying to maximize the yield on its equity tranche, Magnetar surely influenced the makeup of its CDOs: it had every interest in building structures with high equity yields, and those structures by their nature were likely to be pretty risky. What’s more, Magnetar threw so much money at this trade that it helped to fuel the entire subprime bubble.

But between the disclosures and the long-equity part of the Magnetar Trade, it’s becoming pretty clear that this particular Abacus deal is significantly more egregious than anything Magnetar did. If you take all of the Magnetar deals and put them together, then they become bad by dint of sheer size. But if you’re a lawyer looking to nail a financial market professional for doing something wrong, then you’d do as the SEC did, and pass over both Magnetar and Paulson on your way to Goldman Sachs.

Goldman, after all, has to resort to saying, with a straight face, things like this:

The offering documents contained nothing materially false or misleading about ACA’s role, and no reasonable investor would have needed disclosures describing the participation of Paulson…

The fact that Paulson was unknown to ACA – which, as of May 31, 2007, had 26 CDOs valued at $17.5 billion under management – demonstrates that the fact of Paulson’s involvement would not have been material. Nor is there any evidence that IKB or ABN knew of Paulson at the time or would have changed their investment decisions one iota had they fully understood his involvement…

Similarly, the fact that ACA may have perceived Paulson to be an equity investor is of no moment.

None of this is plausible on its face. Goldman repeatedly says that ACA was a “reasonable investor” in the deal, and ACA clearly needed some disclosures with respect to Paulson’s role. Not because of who he was, but because of his reason for doing the deal in the first place. They thought he was with them, on the long side; instead, he was against them, on the short side. That knowledge, no matter who he was, would surely have greatly affected the back-and-forth between the two sides when they worked out a group of names which was mutually acceptable. Either you’re working with someone, or you’re negotiating against them. The difference is crucial, and Goldman should have made it clear.


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

“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”

Tourre actually mentions the above and then he claims to remember nothing about it. This part strikes me as very suspicious. Then they dance around saying this didn’t amount to an equity position which doesn’t make sense.

The whole response smacks of arrogance and seems dismissive of the SEC. No wonder they were blindsided as they thought the SEC would have been dazzled by it and would quietly drop the investigation.

The Magnetar statements are also puzzling. The firm seems to be going to great lengths in their PR effort even though there are no charges against them. They are also portraying themselves as simple, market-neutral arbitrage traders who just got lucky on their shorts, when their terrible longs evaporated. A HF wagering billions without having a directional bias and hoping for their longs to pay off in 6-8 years is just too bizarre to believe.

Posted by mister_x | Report as abusive


You write: “It’s a bit weird, that Goldman would leave itself with the 45-50% tranche of the Abacus deal, but that seems to have been the case.”

Might this have been due to the fact that, by March 2007, the US house price/subprime bubbles had noticably burst, which meant there was a shortage of suckers likely to be taken in by Goldman’s deceitfulness?

Just like RBS Greenwich at that time, was Goldman having to start eating its own (poisoned) cuisine? see this article

Posted by IanFraser | Report as abusive

“That doesn’t mean that the Magnetar Trade wasn’t all sweetness and light…”

I think you mean “was” there.

Posted by klhoughton | Report as abusive

Magnetar’s ability to pay its protection premiums with the earnings from its equity is funny as hell. Which is why the zero trigger on those CDOs was key. And why it is not surprising that the banks could not find suckers to unload that stuff on.

Posted by rootless | Report as abusive

Felix, you miss the beauty of the Magnetar trade. What Paulson did was risky if the market had not collapsed so soon, he would have been stuck paying those massive premiums. What Magnetar did was fund the short premiums with their equity purchase so that a delay in market crash would not have cost them.

Posted by rootless | Report as abusive

It’s true that this Abacus deal had some unique characteristics that made it worse than the Magnetar deals. But it’s still an open question where the SEC is going to draw the line between acceptable and unacceptable disclosure. After all, even the Magnetar deal could have had better disclosure about the influence of the equity on portfolio selection, right? And you could extend this logic to almost any CDO. After all, the interests of the equity and senior investors are never perfectly aligned in a CDO. The equity usually wants higher-spread names because they get the benefit of the excess yield, whereas the senior want the safer names, because they want to keep getting a tiny reliable spread. So are the SEC going to go after every CDO deal in which the equity may have influenced portfolio selection and this wasn’t explicitly disclosed in the offering documents?

Posted by o_nate | Report as abusive


Goldman’s defense is based on Caveat Empor, that the buyers of the ABACUS synthetic CDO should have performed due diligence. The central question is how much DD should an investor have to do? Should he have to go back to the FICO score, stated income, etc, etc for every homeowner involved? Remember, these synthetic CDO’s had a high percentage of traunches with AAA ratings. If I buy a bond with a AAA rating, I assume that S&P or Moody’s has done the due diligence, they are the experts. If I was a fund manager for let’s say Thrivent Financial for Lutherans, and I go to my boss about a CDO, and I say, “yes, I know this traunche we are buying is AAA, but I want to spend several days of my time doing the same due diligence the bond rating agency has supposedly already done.” And the boss is going to ask why I am wasting my time duplicating something S&P has already done, and then he would ask me if I think I am smarter than the ratings agencies.

Somebody tell me why that fund manager should have to do extensive due diligence on a AAA CDO.

Posted by randymiller | Report as abusive

Whether ‘its nobler in the mind to lose $90Million on paper or to earmark it as lunch money for Twitterers muddying up the issue, that is the question.

Posted by HBC | Report as abusive

sorry, klhoughton. Fixed.

So the idea is that the Abacus deal is worse than Magnetar because at least Magnetar took a long position in the equity tranche while Paulson was just flat out short the reference portfolio. I’m not sure how that makes it better. After all, if you own a piece of the equity tranche and are short the reference portfolio, your equity position essentially funds your CDS premiums. That also creates a huge incentive to load up the deal with crappy securities as it will enhance the yield on your equity tranche position so that until the thing implodes and you get the big CDS payout you are compensated nicely in the meantime. I may very well be missing something, though..

Posted by ajkurki | Report as abusive

This may be a little bit of a sidebar, but I am puzzled how the value of the ABACUS CDO went to zero on residential mortgage bonds as the original device.

For example, let’s say that in 2007 I packaged 5 mortgages into an RBMS, those mortgages each being 0% down, low FICO scores, interest only, and the homeowner was lying about his income. In my example, each of the mortgages was $200,000, so the total RBMS book value is $1 million. This is a worst case scenario.

Then, in the example, 10 identical packages were bundled into a CDO, total book value of 10 million dollars.

If everybody defaulted, and that market lost half its value, the CDO should still have half its value, 50 cents on the dollar. OK, discount a little more for the foreclosure costs, but there would still be some value there.

And if a synthetic CDO references my CDO, and my CDO is worth 50 cents on the dollar, how does it go to zero?

If the owners of the CDO were trying to dump it at any price to cover positions, I could see some panic selling where they would take 10 cents on the dollar, but all the way to zero?

Posted by randymiller | Report as abusive

Certain lower-rated tranches, which are designed to carry a higher yield and absorb the first losses, certainly went to zero. The highest tranches probably had some recovery value.

Posted by ajkurki | Report as abusive

randymiller, if the synthetic references not your CDO but rather just the BBB tranche of your CDO, then it can easily go to zero.

The executive leadership teams at our nation’s largest banks and financial insitutions are responsible for everything that happens or fails to happen within their domains. If firms such as Goldman Sachs cannot ethically manage their affairs in detail (including the actions of their non-principal employees), then we should dissolve Goldman as “too big to manage.” The nation is exhausted from the excesses on Wall Street, and if the leadership at Goldman Sachs cannot fulfill their duties and responsibilities, then let’s proceed now with breaking them up for the good of the banking industry, as well as the nation. Again, the senior mangement teams at our nation’s banks and insurance companies are responsible for everything that happens or fails to happen within their domains, and quibbling about that responsibility is outrageous. More at: -and-financial-services-are-in.html

Thank you for the opportunity to comment…

Posted by mckibbinusa | Report as abusive

One question I don’t understand is how was 79% of the ABACUS 2007-AC1 transaction was rated AAA when it was a synthetic CDO (basically just a complex structured note, as I understand it)? Did/do participants on opposite sides of the transaction have to overcollateralize the risk to get the high rating or is it done with some other form of alchemy?

Also, while I’d never want to put myself in the position of defending the bloodsucking vampire squid, why isn’t their more outrage directed at the moron fiduciaries at IBK and ABN that were buying these sorts of garbage securities? As a professional investor, one rule of survival I’ve considered important is to grab my wallet and run whenever somebody shows me an incredibly complicated transaction that can barely be understood on a simple level – any sentient adult ought to know that such deals are designed to fleece . . . . . .

Posted by Anarchus | Report as abusive

A CDO isn’t the entire entity (e.g., ABACUS 2007-AC1), but rather the various notes issued by that special purpose vehicle. Additionally, not all the notes issued are identical, as some “tranches” are riskier (absorb all of a certain percentage of the first losses), while some tranches are much safer and can only lose value if a very substantial portion of the reference securities decline in value.

Posted by ajkurki | Report as abusive

I understand how the CDO tranches worked to get the AAA ratings on the highest preference notes, because there were actual fixed income securities in the special investment vehicle.

What I don’t understand is how the high AAA credit ratings were obtained on “synthetic” CDOs where the returns were based on mathematical calculations from market prices rather than actual divvying up of physical cash flows from real securities.

Posted by Anarchus | Report as abusive

A synthetic CDO is designed to replicate the payoff stream of a given reference portfolio, so really the ratings of that portfolio of cash bonds should be the same as the synthetic CDO.

Posted by ajkurki | Report as abusive

“None of this is plausible on its face.”

Jesus, Felix, you might want to tone down the chutzpah. You have no clue what arguments are implausible on their face, and you’re way out of your depth on matters of securities law. You’re in no position to be judging the strength/weakness of the legal arguments in this case. About half of your posts on the Goldman case so far have been completely wrong about one thing or another, so it’s probably about time for you to start saying, “Okay, I don’t really know what I’m doing here, but here are my thoughts…”

Posted by jhedges | Report as abusive

“Somebody tell me why that fund manager should have to do extensive due diligence on a AAA CDO.”

Well, why don’t you tell me why anyone should pay a fixed income fund manager if they’re not going to do due diligence (or more accurately, credit work)? If they only want to buy high yielding triple-A bonds, they could just automate the whole process.

Posted by GingerYellow | Report as abusive

GingerYellow – excellent point – “then fire the fund manager if their sole value add is to read the ratings label”

re Magnetar’s pleadings:
“Magnetar’s strategy was in essence a capital structure arbitrage.”
Eh? I thought a capital structure arbitrage would work the reverse of Magnetar, ie short the equity, long the debt. (unless i wanted to ‘bribe’ the sponsor by buying the toxic basement to influence construction of the upper tiers)

Posted by essorkm | Report as abusive

Am I missing something? I read the 2007 ABACUS offering and in the conflict of interest section it explicitly states,
“The Initial Purchaser, the Protection Buyer, the Basis Swap Counterparty, the Collateral Put Provider, the Collateral Disposal Agent and their respective affiliates may hold long or short positions with respect to Reference Obligations and/or other securities or obligations of related Reference Entities and may enter into credit derivative or other derivative transactions with other parties pursuant to which it sells or buys credit protection with respect to one or more related Reference Entities and/or Reference Obligations.”

Again, am I missing something. Seems that GS made all the appropriate disclosures…

Posted by mrbuc | Report as abusive