All the Devils Are Here

by Bethany McLean; Joe Nocera

Once again, Goldman had to push hard to sell the deal. Finally, though, Goldman was able to sell about $300 million of Timberwolf securities to Bear Stearns Asset Management. The firm sold another $78 million—at a sizable discount—to an Australian fund called the Basis Yield Alpha Fund, which at the time had only $500 million under management. According to a $1 billion lawsuit Basis later filed—a suit whose central claim is that the Timberwolf purchase forced Basis into insolvency—Goldman told Basis that the market was stabilizing. And while the Timberwolf prospectus states that Goldman owned equity in Timberwolf, the Senate Permanent Subcommittee would later highlight that Goldman also had a substantial short position. “Goldman was pressuring investors to take the risk of toxic securities off its books with knowingly false sales pitches,” said Basis’s lawyer. Goldman called the lawsuit “a misguided attempt by Basis, a hedge fund that was one of the world’s most experienced CDO investors, to shift its investment losses to Goldman Sachs.” One fund manager who knows Basis has a different take: “Dumb money,” he says.

  Within a year, Timberwolf’s triple-A securities had been downgraded to junk, as the WaMu option ARMs defaulted. The Goldman trader responsible for managing the deal later characterized the issuance of Timberwolf as “a day that will live in infamy.” Tom Montag put it more bluntly. “Boy that timeberwof [sic] was one shi**y deal,” he wrote on June 22, 2007. Once again, Goldman insists that it lost hundreds of millions of dollars on the Timberwolf deal, but to the extent that the deal provided a way for Goldman to exit or hedge existing positions, the firm lost less than it would have otherwise.15

  • And then there was Abacus 2007-ACI, the most infamous of all the Goldman synthetic CDO deals. Nearly three years after the deal was completed, the SEC would charge Goldman with fraud, alleging that the firm made “materially misleading statements and omissions” in connection with the deal. Goldman heatedly disputed the SEC’s charges at first, but ended up settling the case for the record sum of $550 million and conceding that the marketing materials were “incomplete.” But in truth, the legal issues were far from the most disturbing thing about Abacus 2007-ACI.

  It began with John Paulson, then a little-known hedge fund manager, who along with Andrew Redleaf, Michael Burry, and a handful of others, had been painstakingly buying credit default swaps on subprime mortgage-backed securities. Paulson and his staff were convinced that the entire mortgage market was poised to collapse. Their analysis, in retrospect, was prescient. As a Paulson employee wrote in January 2007, “[T]he market is not pricing the subprime RMBS wipeout scenario. In my opinion this situation is due to the fact that rating agencies, CDO managers and underwriters have all the incentives to keep the game going, while ‘real money’ investors have neither the analytical tools nor the institutional framework.” Anticipating that “wipeout scenario,” Paulson was seeking to do something that would have a big potential payoff. He wanted to make an industrial-sized short by betting against all the triple-A tranches of a single synthetic CDO—a CDO, in fact, that he would secretly help construct. In other words, he would make money if homeowners couldn’t pay their mortgages—and to improve his odds, he was going to, in effect, select which homeowners he thought were least likely to pay.

  It was an astonishingly brazen idea—like “a bettor asking a football owner to bench a star quarterback to improve the odds of his wager against the team.” That was the description Scott Eichel, a Bear Stearns trader, gave to Gregory Zuckerman, the Wall Street Journal reporter whose book The Greatest Trade Ever documented Paulson’s audacious short. Eichel explained to Zuckerman that when Paulson broached his idea with Bear Stearns, it said no. “[I]t didn’t pass the ethics standards,” said Eichel. It didn’t pass Bear Stearns ethics standards? The same Bear Stearns that had created some truly terrible subprime securities without batting an eyelash? Yet Goldman Sachs had no such qualms.

  Paulson knocked on Goldman’s door at a fortuitous moment. The firm had begun thinking about “ABACUS-rental strategies,” as Tourre described it in an e-mail. By that, he meant that Goldman would “rent”—for a hefty fee—the Abacus brand to a hedge fund that wanted to make a massive short bet. Paulson’s idea fit perfectly.

  Paulson paid Goldman $15 million to rent the Abacus name. The buyers of the CDO—the longs on the other side of the Paulson short—assumed it was a deal instigated by Goldman, since Abacus was a Goldman platform. They had no idea that Paulson was helping to select the securities that would make up the deal. Indeed, as the deal was nearing completion, the Paulson team decided to throw out mortgages originated by Wells Fargo. Wells Fargo mortgages, after all, might actually perform. Goldman’s failure to disclose Paulson’s involvement in selecting the securities in its marketing material for the transaction became the heart of the SEC’s case against the firm.

  (According to one person familiar with the deal, Goldman even contemplated keeping the short position for itself instead of giving it to Paulson, who was not considered an important client. The irony is rich: had Goldman kept the short position for itself, it would have double-crossed Paulson, but the SEC would have had no case.)

  There were no clean hands here. In renting the Abacus platform and helping to select the referenced securities, Paulson was doing something that may have been perfectly legal, but was awfully sleazy. He wound up shorting most of the $909 million super-senior tranche. The rating agencies were cooperative, as always, even though the Abacus deal was specifically stocked with securities that had been chosen in the expectation that they would fail. Eric Kolchinsky, the Moody’s analyst who oversaw the rating process, later testified that he hadn’t known about Paulson’s involvement and that it was “something that I personally would have wanted to know.” He added, “It just changes the whole dynamic of the structure, where the person who’s putting it together, choosing it, wants it to blow up.” But this was a lame excuse. If there was one party with a duty to do its own due diligence on the securities Abacus referenced, surely it was the rating agencies.

  The CDO manager that was supposed to be choosing the securities, a firm called ACA Management, took its fees and appeared to look the other way—exactly what Goldman hoped it would do. E-mails show that one CDO manager had even turned the deal down “given their negative views on most of the credits that Paulson had selected,” as Tourre wrote. (The SEC claimed that ACA didn’t understand that Paulson was going to short the deal, which is a little hard to believe.) ACA also invested $42 million in the securities, and its insurance arm took the other side of the Paulson bet by guaranteeing the $909 million in super-senior tranches.

  The final counterparty was an Abacus veteran: IKB. IKB was no lamb being led to slaughter. It had bragged incessantly about its expertise in the CDO market and, according to a lawsuit later filed against it by the French bank Calyon, was trying to off-load its own bad deals onto others. In June 2007, IKB also created a structured investment vehicle called Rhinebridge. Rhinebridge, like other SIVs, issued debt that it then used to buy mortgage-backed securities and CDOs like Abacus. The debt issued by Rhinebridge, which was rated triple-A, was bought by, among others, King County, Washington, which managed money on behalf of one hundred other public agencies. This was money used to run schools and fix potholes and fund municipal budgets. Rhinebridge was wound down in the fall of 2008, with its investors getting fifty cents on the dollar. In a lawsuit, King County alleged that IKB created Rhinebridge “for the purpose of moving investment losses off of its own balance sheet.” For all of Goldman’s later claims that it dealt only with the most sophisticated of investors, the fact remained that those investors could be fiduciaries, investing on behalf of school districts, fire departments, pensioners, and municipalities all across the country. It was their money, at least in part, that was funding the CDO games Wall Street was playing.

  There was another problem with the “sophisticated investor” defense. These deals were so complicated that in many cases nobody understood the risks, not even the underwriter. Yet investors—eve
n sophisticated investors like IKB—were buying deals like Abacus for a simple reason: they didn’t want to lose money. Triple-A-rated securities were supposed to be the closest thing an investor could get to a risk-free investment. If Goldman knew that a triple-A rating no longer meant what it once had—and that these complex securities carried far more risk than their ratings implied—did it really have no responsibility to say anything? Shouldn’t there have been a point at which Goldman just said no? If Paulson’s bet paid off, it would happen because millions of Americans were losing their homes. Wasn’t that worth thinking about before deciding to go through with the Abacus deal? In 2004, Scott Kapnick, who had headed Goldman’s investment banking department, had said to Fortune, “The most powerful thing we can do is say no.” But by 2007, Kapnick, like many other senior bankers, had left the firm.

  On March 7, 2007, Tourre sent an e-mail to his girlfriend. “I will give you more details in person on what we spoke about but the summary of the US subprime business situation is that it is not too brilliant… According to Sparks, that business is totally dead, and the poor little subprime borrowers will not last so long!!! All this is giving me ideas for my medium term future, insomuch as I do not intend to wait for the complete explosion of the industry.”

  That same day, the Goldman Sachs Firmwide Risk Committee heard a presentation from the mortgage desk. According to a summary of the meeting, the first bullet point read, “Game Over—accelerating meltdown for subprime lenders such as Fremont and New Century.” The second bullet point: “The Street is highly vulnerable, potentially large exposures at Merrill and Lehman.”

  On June 24, Gary Cohn, at the time Goldman’s co-chief operating officer, sent an e-mail to Viniar and several others, noting both the big losses Goldman was taking on the mortgage securities it had been unable to “distribute” and the even bigger gains it was booking from its short position. Viniar’s response: “Tells you what might be happening to people who don’t have the big short.”

  Later, the Senate Permanent Subcommittee would charge Goldman with making a fortune—$3.7 billion—by betting against its customers when it knew the market was going to fall apart. But that’s not really what happened. The huge gains Goldman made from its short position in 2007 were offset by substantial losses from the securities it couldn’t get rid of. Indeed, the firm made less money than it might have, because at certain points during the meltdown, most notably in the spring of 2007, Goldman covered its short position. It didn’t envision the “wipeout scenario,” as Paulson had. Rather, it was trying to figure out what the market was doing and stay one step ahead of it. Overall, Goldman’s mortgage department made $272 million in the first quarter of 2007, lost $174 million in the second quarter, made $741 million in the third quarter, and made $432 million in the fourth quarter. In total, Goldman’s mortgage department made $1.27 billion in 2007, a big number, obviously, but not even close to the $4 billion John Paulson made. “Of course we didn’t dodge the mortgage mess,” Goldman CEO Lloyd Blankfein wrote in the fall of 2007. “We lost money, then made more than we lost because of shorts.” True enough.

  Other firms besides Goldman were also trying to dump their exposure onto buyers who hadn’t figured out that the ratings had become degraded. Other firms also sold synthetic CDOs while keeping a short position. But Goldman was unquestionably better at it than its competitors. What Goldman Sachs really did in 2007 was protect its own bottom line, at the expense of clients it deemed disposable, in a conflict-ridden business that maybe—just maybe—the old Goldman Sachs would have been wise enough to stay away from.

  In all the subsequent frenzy over who did what to whom in the synthetic CDO market, a series of deeper, more troubling questions tended to get overlooked. One was this: What, exactly, was the point of a synthetic CDO? It didn’t fund a home. It didn’t make the mortgage market any better. It was a zero-sum game in which the dice were mortgages.

  “Wall Street is friction,” said Mark Adelson, the former Moody’s analyst. “Every cent an investment bank earns is capital that doesn’t go to a business. With an initial public offering, you get it. But with derivatives, you can’t tie it back. You could argue that at least it’s not hurting things, and that was a compelling rationale for a long time.” He concluded, “We may have encouraged financial institutions to grow in ways that do not directly facilitate or enhance the reason for having a financial system in the first place.”

  If only that were the worst of it. But it wasn’t. The invention of synthetics may well have both magnified the bubble and prolonged it. Take the former first. Synthetic CDOs made it possible to bet on the same bad mortgages five, ten, twenty times. Underwriters, wanting to please their short-selling clients, referenced a handful of tranches they favored over and over again. Merrill’s risk manager, John Breit, would later estimate that some tranches of mortgage-backed securities were referenced seventy-five times. Thus could a $15 million tranche do $1 billion of damage. In a case uncovered by the Wall Street Journal, a $38 million subprime mortgage bond created in June 2006 ended up in more than thirty debt pools and ultimately caused roughly $280 million in losses.

  As for prolonging the bubble, synthetics likely did it in two ways. Firms were much more willing to buy and bundle subprime securities from some of the worst originators knowing they could use a synthetic CDO to hedge any exposure they might be stuck with. Would Goldman have sold over $1 billion of Fremont mortgages to investors in early 2007 if it hadn’t been able to enter into credit default swaps to hedge some of its own resulting exposure to Fremont? Without the means to off-load these exposures, investment firms would likely have been more cautious—and shut off the spigot sooner.

  Secondly, selling the equity in the CDO, the riskiest piece, required finding buyers willing to take that risk. There weren’t that many to begin with, and once they had enough equity risk on their books and stopped buying, the market for mortgages would have naturally wound down.

  But around 2005, some smart hedge funds began to realize that there was a compelling trade to be made by buying the equity in a CDO while shorting the triple-As. If the mortgages performed, the return offered by the equity pieces, which could be upwards of 20 percent, more than covered the cost of the short. And if the mortgages didn’t perform? Then the short position would make a fortune. It was a classic correlation trade. It was also practically foolproof.

  The arrival of this trade may have been the final bit of juice that the market needed to keep from running out of gas. No longer did the underwriter have to find buyers willing to take on the equity risk. Instead, buyers of the equity slice could not have cared less about the risks in that portion of the CDO. If the equity made money, they made money. If the equity lost money, they made even more money. Suddenly, the equity portion was a very easy sell.

  This trade gained popularity just when it looked on the ground like the subprime madness was grinding to its inevitable end. Instead, the business kicked into one last crazed frenzy, as subprime originators handed out mortgages to anyone and everyone.

  “Equity is the holy grail of CDO placement,” wrote Lang Gibson, the Merrill Lynch CDO researcher. “The compelling economics in the long ABS correlation trade [buying the equity while shorting more senior tranches] will propel the mezz CDO market forward, no matter the evolution of fundamentals in residential mortgage credit, in our view.” Which is exactly what happened.

  In January 2007, Tourre sent another e-mail to his girlfriend. “Work,” he wrote, “is still as laborious, it’s bizarre I have the sensation of coming each day to work and reliving the same agony—a little like a bad dream that repeats itself…. When I think that I had some input into the creation of this product (which by the way is a product of pure intellectual masturbation, the type of thing which you invent telling yourself: ‘Well, what if we created a ‘thing,’ which has no purpose, which is absolutely conceptual and highly theoretical and which nobody knows the price?’). It sickens the heart to see it shot down in mid-fl
ight…. It’s a little like Frankenstein turning against his own inventor….”

  Of course, the one thing that was neither conceptual nor theoretical was the losses. They were all too real, and in 2007, as winter turned to spring, they were coming.

  19

  The Gathering Storm

  Questions about who owns the risk—it’s spread out all over the world in various formats including repackaging vehicles. Not that obvious to find out who is feeling the pain.

  —Dan Sparks e-mail, March 1, 2007

  On Friday, March 2, 2007, a man named Ralph Cioffi, who ran two hedge funds at Bear Stearns that had some $20 billion invested in asset-backed securities, held a small, impromptu meeting in his office. Matt Tannin, who managed the two funds with him, was there, as was Steve Van Solkema, a young analyst who worked for the two men and another partner in the funds. They had gathered to discuss the deteriorating market conditions. The week had opened with a drop in the stock market of more than 400 points, the largest one-day decline since the aftermath of 9/11. Cioffi described February as “the most treacherous month ever in the market.” They talked about the plunge in value of the riskier tranches of the ABX index. Even some of the triple-A—the triple-A—were showing a strange wobbliness. That wasn’t supposed to happen—ever. The men were anxious.

  On paper, their two hedge funds hadn’t performed that badly: one fund was down a little; the other was up a little. But it had suddenly become difficult to obtain prices on the securities they owned, so they couldn’t be sure what their funds were truly worth. Plus, they’d often told investors that the funds operated like a boring, old-fashioned bank—they were supposed to earn the difference between their cost of funds (a good chunk of which were provided through the repo market) and the yield on the super-safe, mostly triple- and double-A-rated securities that they owned. Investors expected fairly steady, low-risk returns. Any losses, no matter how small, could spook them. The Bear team had made money on short positions they had placed on the ABX, but the volatility was worrisome. Because the higher-rated securities were supposed to be nearly riskless, the Bear Stearns hedge funds were highly leveraged: only about $1.6 billion of the $20 billion was equity. The rest was borrowed. Earlier in February, they’d started to get margin calls, meaning that their lenders were demanding more collateral. They’d met the margin calls, but their fears had not abated.