Although Goldman made large profits on its mortgage shorts, it could have made much more. But top management—Viniar, Cohn, and Lloyd Blankfein, the CEO—decided against truly massive short bets. They loved beating their competitors, but they wanted to stay safe, preferring to lock in smaller immediate profits as protection against the risks posed by the fear, uncertainty, and destruction cascading throughout the entire global financial system as the crisis widened. One of Goldman’s traders, Josh Birnbaum, later lamented that his desk could have made more money than John Paulson’s fabled “Greatest Ever Trade” if only management had let him.9 But this would have required making significant investments in buying short positions and also postponing the booking of profits from them, a strategy that carried with it the risk that investors would panic if Goldman reported losses in 2007 or 2008.
As a result of Goldman’s foresight and caution, 2007 was the best year ever for its mortgage business. By the end of the third quarter, and as Table 1 below illustrates, mortgage revenues were nearly equal to full-year 2006 revenues, the previous high. The revenue jump came from the shorts, which were concentrated in the structured products desk. The losses in residential mortgages came from Goldman’s decision to mark down its remaining long positions.
To be fair, when Goldman managers referred to their mortgage business as “modest”, they were telling the truth. Full-year 2007 mortgage revenues were about $1.2 billion. But that is what makes Goldman’s performance so impressive; they understood that in the new financial world of derivatives and massive leverage, even a modest business could suddenly generate catastrophic, multibillion-dollar losses if they were caught unprepared. Contrast that with, say, Citigroup’s behaviour, not to mention Jimmy Cayne’s. Even though Citigroup had a much larger position—more than $50 billion of high-risk home mortgages on their books—senior management didn’t even know they had a problem until the market cratered.
So full marks to Goldman for cleverness. But their strategy wasn’t just to short the mortgage market, because sometimes it was better to sell off the riskiest assets instead of paying for insurance on them. Once again, this was accomplished with impressive efficiency, but also with a near-total disregard for ethics, integrity, and their customers’ welfare. Whether their behaviour meets the standard of criminal fraud may never be entirely known, but without question it was really, really greasy.
In reading what follows, you might measure Goldman’s conduct against its pretentious corporate declaration that “integrity and honesty are at the heart of our business. We expect our people to maintain high ethical standards in everything they do.”10
Clearing the Shelves
IT WAS ONE thing to decide to sell off risky assets, but quite another to find people still willing to buy them in late 2006 and 2007. Clever people were already out of the question; only fools would do. In December 2006, Fabrice (“Fabulous Fab”) Tourre vetoed a list of sales prospects because it was “skewed towards sophisticated hedge funds . . . most of the time they will be on the same side of the trade as we will, and . . . they know exactly how things work.”11
Luckily there were fools still to be found. The sales team started circulating “axes” (meaning sales priorities) in December 2006. A few months earlier, a trader had mentioned that “sales people view the syndicate ‘axe’ e-mail we have used in the past as a way to distribute junk that nobody was dumb enough to take first time around.”12
“The key,” according to a March 2007 memo, was “getting new clients/capital into the opportunity quickly.” In effect, dumb money, or as another note put it, “non-traditional buyers”. A salesman pushing one of Goldman’s most toxic offerings to a Korean client thought he could expand the sale, but wanted a better commission “as we are pushing on a personal relationship” [i.e., I only screw my friends if I am paid well for it]. Sparks agreed.13
One of the earliest house-cleaning operations was a $2 billion synthetic CDO called Hudson Mezzanine Funding 2006-1. The sales presentation is almost completely misleading. Goldman’s intent, it said, “is to develop a long term association with selected partners” by creating “attractive proprietary investments.” Goldman, it claimed, had “aligned incentives with those of the investors.” The presentation materials stated that “Goldman Sachs CDO desk pre-screens and evaluates assets for portfolio suitability . . . and reviews individual assets in conjunction with respective mortgage trading desks.” True, in a sense—because it was the CDO desk that was choosing the assets they wanted to get rid of.14 “Assets sourced from the Street. Hudson Mezzanine Funding is not a Balance Sheet CDO.” This last statement, however, was a lie. It was made to reassure investors who had already learned to be sceptical of banks unloading bad inventory.15
Which was exactly what Goldman was doing. When the deal was executed, cheers resounded through the trading group—they had engineered a big reduction of risk and booked $8.5 million in earnings. A few months later, after shedding more risk through another sale, Sparks commended the team for having “structured like mad, and travelled the world, and worked their tails off to make some lemonade out of some big old lemons.”16
But Goldman had even more tricks to play on the Hudson investors. Virtually all the Hudson securities were severely downgraded in October 2007, which qualified as a liquidation event. As the liquidation agent, Goldman was supposed to sell the securities at the best available price and supervise the settlements with investors. Instead, Goldman told investors that it would delay liquidation until it could find a more experienced agent (!). The price of the securities continued to plummet; instruments worth 60 cents on the dollar in October were worth only 16 cents by January. An asset manager from Morgan Stanley3 wrote to Goldman in February, “No good reason to wait other than to devalue our position. It’s a shame . . . one day I hope to get the real reason why you are doing this to me.” Well, the real reason they were doing it to him was that Goldman had shorted the securities, so every additional drop in price transferred more wealth from the investors to Goldman.17
And then there was Timberwolf, another synthetic CDO in the amount of $1 billion—the one that was immortalized by Senate hearings, and that prompted a suggestion for paying “ginormous” sales credits. Here is an e-mail trail prompted by a note from sales saying they had just closed a sale on part of Timberwolf at 65 percent of its original value.
TRADER: This is worth 10. It stinks . . . I don’t want it in our book.
SALES: It is not that bad.
TRADER: Cds mkt thinks that deal is one of the worst of the year . . . hopefully they r wrong.18
Goldman management was happy to get rid of it. A 28 March 2007 e-mail to the sales syndicate reads: “GREAT JOB . . . TRADING US OUT OF OUR ENTIRE TIMBERWOLF SINGLE-A POSITION.”19
In June, with $300 million of Timberwolf assets still to be sold, Tom Montag, Dan Sparks’s boss, made the comment quoted later at Senate hearings, writing, “boy that timberwolf was one shitty deal.”20 But, of course, Goldman kept on selling it.
Later, in September 2007, Tourre asked for a specific Goldman deal that a potential client could use as an example for a hedging strategy. Two traders exchanged e-mails on the specifics:
TRADER 1: which class, exactly
TRADER 2: Not sure, a class that went from near par in Jan to around 15 now.
TRADER 1: Well [Timberwolf] didn’t exist until 3/27/07. . . . Here is the basic history. . . .
3/31/07 94-12
4/30/07 87-25
5/31/07 83-16
6/29/07 75-00
7/31/07 30-00
8/31/07 15-00
Current 15-00
TRADER 2: 3/27—a day that will live in infamy.21
But Tourre’s question prompts a double take. Why would a customer sales guy want to advertise a product with such dreadful performance?
Why, indeed. Well, we don’t know the rest of that particular conversation, but there is ome very logical explanation for what Fabulous Fab was doing. He, and Goldman, wa
nted to do business not only with fools, but also with clever people, that is, people who were looking for things to short. People who even wanted Goldman to help them construct securities specifically tailored to fail. People like John Paulson.
The SEC started investigating the Paulson short in late 2008, but not until 2009, with the publication of Greg Zuckerman’s The Greatest Trade Ever, was it widely known that Goldman had built a deal just for Paulson to short. ABACUS 2007-AC1, known as AC-1 within Goldman, mostly comprised a menu of bonds chosen by Paulson for being particularly bad, in a market where bad was average.
The Paulson Trade
FUND MANAGERS HAD been shorting the housing market long before John Paulson; we saw that Morgan Stanley started in 2004. But as Morgan Stanley learned to its dismay, market bubbles can outlast bearish investors’ resources, and as time passed and the bubble and its inevitable collapse became more obvious to insiders, it became more expensive to short housing.
Paulson ran several hedge funds and had the resources to lose money for a long time. But, convinced as he was of a coming “subprime RMBS wipeout”, even he worried about the opacity and complexity of mortgage securities. So Paulson explored whether an investment bank might custom-design one, so that he could know and even choose exactly what went inside. Even Bear Stearns, not exactly a paragon of ethics in mortgage dealings, turned Paulson down because his request seemed “improper”.22 As we saw earlier, Bear Stearns behaved in an extraordinarily unethical way, but although they weren’t concerned about ethics, they did appear to fear potential prosecution for fraud. Happily for Paulson, Goldman seemed to have no such scruples.
Fabulous Fab Tourre designed the deal, communicating with investors and overseeing the preparation of marketing materials. To make the deal easier to sell to the long investors, aka suckers, Tourre needed an independent manager supposedly responsible for selecting the investment assets. In internal correspondence Goldman traders worried that such a manager might not “agree to the type of names Paulson want to use” or would refuse to “put its name at risk . . . on a weak quality portfolio.” ACA Management LLC, which had worked as manager on other Goldman deals, got the assignment.23
ACA was never told the real purpose of the transaction. They were aware that Paulson would be involved in the deal, but not that his real goal was to bet on its failure. Paulson provided Goldman with a list of 123 bonds he wanted to short; the list was supplied to ACA, and meetings were held between ACA, Paulson, and Goldman. Although ACA was occasionally puzzled by the names Paulson recommended, or the strong ones he vetoed, an agreement was worked out on a ninety-bond portfolio, which included fifty-five of Paulson’s original names. At one point in the discussions, Tourre e-mailed a colleague at Goldman, “I am at this aca paulson meeting, this is surreal.”24
All the materials strongly emphasized ACA’s role; none mentioned Paulson’s. The marketing emphasized the deal’s attractions for the long investor. The heart of the sales book, for example, is an eighteen-page statement on the credit selection skills of ACA. Sample quotes: “Asset selection . . . premised on credit fundamentals” . . . “alignment of economic interest with that of investors” . . . none of “ACA’s CDOs have ever been downgraded.” And much more in that vein, including a detailed presentation of the very rigorous ACA credit review process.25
The final deal was yet another synthetic CDO, the $2 billion ABACUS 2007-AC1, the majority of whose reference securities had been handpicked for their poor quality. Through the magic of the rating agencies, the majority of the securities were rated AAA. Goldman later said that they lost $90 million on the transaction. This seems unlikely, because Goldman’s own internal documents state that they would not be taking any risk on the deal. But perhaps even Goldman was sometimes gamed by its own employees. Maybe Goldman was unable to sell all of it, was pressured to repurchase some of it, or had some form of liability for losses.26
The targeted customer was IKB Industrie Deutschebanke AG. It was classic uninformed money—a small bank, based in Düsseldorf, with little access to background research. IKB was an eager buyer in CDO markets, and its sponsored hedge fund, Rhinebridge Capital, was one of the first to collapse in the summer of 2007. IKB bought $150 million of the AAA rate notes in April.
Well before the year end, AC-1 was nearly worthless, and Goldman transferred essentially all of IKB’s $150 million to Paulson. Then through complex side deals, some of the long side became a liability of the Royal Bank of Scotland. ACA failed at the end of the year and in early 2008 partial settlements were made, some of which presumably went to Goldman, and from Goldman to Paulson. In August 2008 Royal Bank of Scotland made an $841 million settlement with Goldman on its part of the failed deal, which also would have gone to Paulson.27
The SEC quietly opened an investigation into the Paulson trade in August 2008, and in April 2010 filed a civil fraud complaint in US federal court—its one and only attempt to pursue Goldman Sachs for its behaviour during the bubble and crisis. After declaring that it had done nothing wrong, Goldman settled in July 2010, agreeing to a $550 million fine. In its consent to the judgement, Goldman stipulated that it was “a mistake” to state that ACA had selected the portfolio “without disclosing the role of Paulson & Co. . . . Goldman regrets that the marketing materials did not contain that disclosure.”28
Before moving on to Goldman’s really serious bets with AIG, we should note briefly that, disgraceful as Goldman’s and Paulson’s behaviour was, they were far from alone. Deutsche Bank and several other major investment banks worked with Paulson, as well as with several other large hedge funds, to construct similar deals with similar results. However, it would be surprising to see the SEC go after Deutsche Bank, because since the Obama administration took office, the SEC’s director of enforcement has been Robert Khuzami, who was general counsel to Deutsche Bank for the Americas throughout the bubble. (In a way, though, it could be tidy—he could depose himself, subpoena himself, prosecute himself, and settle with himself.)
As for the other hedge funds that behaved like Paulson, the two best known are Magnetar and Tricadia.
The Magnetar Trade
MAGNETAR IS A Chicago hedge fund that developed a strategy similar to Paulson’s. The brief discussion here is based on an excellent investigative series by the not-for-profit ProPublica news organization, and a more technical analysis in Yves Smith’s book Econned.29
In the Goldman Sachs ABACUS deal just discussed, Paulson did not make any long investments; he merely bought the short side, which sufficed for him to make gallons of money when the securities defaulted. But Magnetar, and sometimes Paulson too in his deals with Deutsche Bank, did something even more clever. They would find, or help construct, really bad deals, and buy the short side of nearly all of them. But then, seemingly paradoxically, they also bought the long side of the so-called equity piece—the lowest-quality part of the whole deal, and the first to fail. Why would they do that?
Because they were handling, in a far more intelligent way, the same problem that poor Howie Hubler had faced at Morgan Stanley. Until the securities actually fail, the guy holding the short side of a synthetic CDO has to keep making those payments to the long investor. That could get expensive. Magnetar (and sometimes Paulson) covered these payments through the extremely high interest rates (often 20 percent per year, sometimes even more) paid by the equity piece. What Paulson and Magnetar realized, and Morgan Stanley didn’t, was that bubbles often last for a long time, but the collapse is usually fast and complete. So the equity piece would keep paying interest, and covering the cost of Paulson’s and Magnetar’s massive short position, until the collapse came. Then, yes, the equity piece would fail first, but everything else would probably fail soon afterwards. This strategy had the additional benefit of disguising their real intentions, at least to naive investors.
Their returns were huge; this was not a small business. An executive from a European bank that managed some of Magnetar’s deals said:
If y
ou told me of a major broker/dealer who had an active CDO underwriting group that *DIDN’T* work with Magnetar . . . that would surprise me. . . . Rinse and repeat. The credits didn’t matter nor really did the managers they contracted. . . . When the math lined up, they would reload the trade.30
In a rare error for Jamie Dimon, JPMorgan Chase was heavily involved in structuring one of these deals, called Squared, and lost money as a result. Magnetar’s behaviour was very similar to Paulson’s with ABACUS. Magnetar helped select the assets that went into Squared, with the intent of betting against it. JPMorgan Chase structured the deal, and then sold off pieces of it without telling its customers that the package had been designed to fail, or that Magnetar was involved. But JPMorgan Chase didn’t sell off enough of the deal, and lost $880 million as a result.
Magnetar pumped out thirty deals, in the $1 billion to $1.5 billion range each. Very rough estimates suggest that this one hedge fund’s short payments may have funded a quarter of the 2006 subprime mortgage securities market. Investment bankers (of the Fabulous Fab sort) collected major fees for helping to structure the deals and matching Magnetar with the fools it bet against. Magnetar was a few clever guys who found a way to turn financial muck and the fools owning it into billions for themselves. If they happened to facilitate another $100 billion to $200 billion of stupid lending and investing, well, that was not their problem.4 Tricadia’s story is very similar.
Inside Job Page 15