Authors: William D. Cohan
Soon after Goldman’s March 2007 board of directors meeting, the Goldman mortgage group began closing one challenging CDO underwriting assignment after another, including those for
Timberwolf, Anderson Mezzanine, and—soon enough, on April 26—ABACUS, which had been such a roller-coaster ride for
Fabrice Tourre. One way the Timberwolf deal got done, according to an internal Goldman memorandum, was because the two hedge-fund managers at Bear Stearns Asset Management, Cioffi and Tannin, bought $400 million worth of the $600 million security—by far the largest chunk—at prices that ranged from just below par (99.7 percent of par) to par.
On Wall Street, Cioffi and Tannin were well-known buyers of squirrelly securities such as CDOs,
CDOs of CDOs (known as CDO squareds), and
synthetic CDOs. Indeed, in October 2006, Goldman had created a $900 million synthetic CDO squared known as
ABACUS 2006 HGS1—a different ABACUS deal than the famous one Tourre worked on—expressly for the two Bear Stearns hedge-fund managers. The security referenced a mix of credit-default swaps on A-rated bonds and synthetic asset-backed securities, “the sweet spot right now” in the market at that time, one trader told
Derivatives Week
in December 2006. Cioffi and Tannin made the Timberwolf deal on March 27, 2007, thanks to the
salesmanship of
Andrew Davilman, then a Goldman vice president. Some six months later—after the value of
Timberwolf had collapsed to around 15 cents on the dollar, Goldman trader
Matthew Bieber referred to March 27 as “a day that will live in infamy.” Meanwhile one of the Bear Stearns hedge-fund investors, who lost all that he had invested, observed, tongue firmly implanted in his cheek: “Nice trade, Ralph.” (According to
Michael Lewis, writing in
The Big Short,
at the same time that Davilman was selling Cioffi most of Timberwolf at par, he was buying insurance from AIG, in the form of credit-default swaps, on behalf of Goldman, as principal, betting that similar CDO securities—although apparently not Timberwolf itself, according to a November 2007 AIG memorandum—would collapse.) Indeed, in Sparks’s March 9 memo—the one where he wrote that the “#1 priority” was to sell “new issues”—he specifically cited Davilman for making a “major contribution” in helping sell the trading “desk’s priorities.” A few days after Cioffi bought the senior tranches of Timberwolf,
Mehra Cactus Raazi was able to sell $16 million of a lower-rated tranche of the same deal. “Great job Cactus Raazi trading us out of our entire Timberwolf single-A position,” an internal memo fairly screamed.
Meanwhile, Tourre was still pounding the pavement trying to sell the ABACUS deal. On March 30, he reported to Sparks that he had been visiting with “selected accounts” during the previous few weeks, many of whom had passed on investing in Timberwolf and Anderson. But there were $200 million of orders—from IKB (apparently having overcome its concern about New Century mortgages being included) and from ACA, the portfolio selection agent. The plan was, he explained, to close the sales of those tranches at the end of the following week and then to try to move the lower-rated tranches shortly thereafter. He urged his colleagues to “steer” accounts “towards available tranches” of ABACUS “since we make $$$ proportionately” when the tranches are sold off.
On April 3, he sought trade approval to sell Paulson & Co. credit protection on $192 million worth of the ABACUS deal, allowing Goldman to book a $4.4 million fee for providing the insurance. He asked Goldman’s credit group to make sure it was OK with the traders. By April 11, though, perhaps because of some push back from the credit group, Tourre was also worried about making sure Goldman maximized its ability to profit from its trading relationship with Paulson & Co., especially after ABACUS closed. He wrote to Cactus Raazi that he needed to ask the Goldman credit department to perform an “updated review” of Paulson “to enable us to put more trades on with these guys” since “it appears” that since the beginning of 2007, Paulson had shorted, through
Goldman, $2 billion notional amount of residential mortgage-backed securities, “which is utilizing most of the credit capacity we have for Paulson.” Tourre explained to Raazi that “[w]e need to be sensitive of the profitability of these trades vs. the profitability of ABACUS—we should prioritize the higher profit margin business with Paulson.” By the next day, it seemed, Tourre had received credit’s approval to do the trades with Paulson, and Raazi booked the trades, much to his dismay—apparently—because he suspected Goldman would get stuck on the losing side. “[S]eems we might have to book these pigs,” Raazi wrote to Daniel Chan, his Goldman colleague.
On May 8, Tourre updated Sparks on the ongoing ABACUS saga, which he referred to as the “short we are brokering for Paulson.” He explained that the “supersenior tranche” of the deal would “most likely” be executed with ACA, through another bank—
ABN Amro, a large Dutch bank—as “intermediation counterparty.” Goldman was to “buy protection” on $1 billion of the security and then Paulson was to short a big chunk of it. (How do they think up these things?) But Tourre was worried that Paulson may have changed his mind about doing the deal as originally conceived. There were now two options for Goldman: one would be a “risk-free” deal where Goldman would make $14 million; the other would make Goldman $18 million but expose the firm to $100 million of risk being long a portion of the deal, although Tourre wrote that he felt confident that risk could be sold at a profit. A week later, with the ABX index rallying, Tourre reported that there was a “90% chance” that ACA and ABN Amro would go through with the deal, but that he was increasingly concerned that Paulson “is starting to get ‘cold feet’ ” on going through with his side of the trade because of the ABX rally. Tourre wanted Sparks’s permission for Goldman to “take down,” or assume the short side of the trade, instead of Paulson, “in order to avoid loosing [
sic
]” the ACA/ABN Amro order.
Two weeks later, Tourre provided the group another update. Paulson had now agreed to his side of the trade—for $1 billion—and ACA/ ABN Amro had agreed to buy $909 million, leaving Goldman with $91 million it was unable to place, although “we are showing this tranche to a few accounts,” he wrote. Finally, the next day, the deal was really done, along the lines Tourre had described the day before. Melanie Herald-Granoff, a Goldman vice president in the mortgage-trading group, wrote to Tourre and
David Gerst, in the structured products trading group: “Fabrice & David—Thank you for your tireless work and perseverance on this trade!! Great job.” By June 5, Gerst was offering up Goldman’s $91 million residual—that piece that neither Paulson nor
ACA purchased—to Bear Stearns Asset Management, or BSAM, at par, or 100 cents on the dollar, with a coupon of Libor plus 0.75 percent. With plenty of its own problems by then, BSAM declined Goldman’s kind offer, leaving Goldman itself on the hook for this piece of the ABACUS deal. After some six months of hard work on the ABACUS deal, Tourre headed to Belgium and then London, in part to visit his girlfriend. “Just made it to the country of your favorite clients [Belgians],” Tourre wrote to Serres on June 13. “I’ve managed to sell a few [ABACUS] bonds to widows and orphans that I ran into at the airport, apparently these Belgians adore synthetic abs [CDO] squared!! Am in great shape, ready to hold you in my arms tonite.”
——
A
T AROUND THIS
time—with Birnbaum’s short bets paying off big—the decision was made to allow his desk to get control of the firm’s growing inventory of those parts of CDOs it could not sell to investors, known as residuals. He and Swenson began to give serious thought to the price it would take to move these residuals off Goldman’s books and into the hands of other investors. Driven in large part by a combination of what the traders were seeing in the market and Jeremy Primer’s models, which kept spitting out lower and lower valuations for the residuals, Birnbaum began to think that the time had come to seriously write down the value of the firm’s CDOs, in order to move them out the door. “Part of it was just a general discipline where we had legacy positions that—frankly—we did not like,” Birnbaum explained. “This is part of the growing purview that our desk had in terms of all the legacy positions. Looking around, we were like, ‘We got to get rid of this shit.’ And when you think it’s worth par, selling it into a ninety-five bid feels pretty bad. But when you think it’s worth seventy, selling it into a ninety-five bid sounds pretty darn good even if you’re taking a five-point loss, right? So, the first thing was just to get the culture of Goldman around that concept. The same percolation upward that was occurring with these short trades was also happening with this valuation question on CDOs.”
This was easier said than done. One Goldman trader remembered a number of critical and contentious meetings at Goldman where different constituents around the firm would submit, in writing, their thoughts about the valuation of the residuals. “I remember looking at one of these sheets,” he said. “There were
huge
differences of opinion on this issue. They were very vehement. There are a lot of senior guys at the firm who have since changed their story very much—and were like, ‘Oh, there’s no fucking way that stuff’s worth that. You guys are crazy.’ Some of the more negative people in our group who were more bearish would really be
pushing the short trades in general. Some of the senior people thought we were a little nuts”—here he did not want to say which senior Goldman executives felt this way. “Ultimately you had the research guys saying, ‘This stuff’s worth below fifty,’ when it was marked at one hundred.” The debate led to a grand compromise. “Fuck it, mark it at seventy,” the Goldman trader said, recalling how the decision got made.
By April, Birnbaum had won the internal debate. Not only did Sparks agree with Birnbaum, he became increasingly concerned about the rapidly declining value of Goldman’s $10 billion mortgage-backed securities portfolio, as more and more home buyers began to default on their home mortgages and the market for the securities tied to them began to cool. He took the laboring oar in convincing the senior Goldman executives that the time had come to move. “We’ve got a big problem,” Sparks told Viniar and Cohn. The decision was made to sell as much of Goldman’s $10 billion portfolio as rapidly as possible, even if the markdowns required to do so were drastic. In the first few weeks of April, the e-mails were flying fast and furious with the mandate to sell the firm’s new “axes,” composed of the residual CDO inventory.
On April 5,
Thomas Cornacchia, head of Goldman’s mortgage sales group, sent around a list of thirty positions, totaling $450 million, that Goldman wanted to sell immediately. “Get this done please,” he ordered, and then added a little bit of competitive verve: “Who is the better salesperson??” Within the hour,
Mehra Cactus Raazi, Goldman salesperson extraordinaire, had sent the list to
Brad Rosenberg, at Paulson & Co., and urged him to take a look. “These are all dirty ’06 originations that we are going to trade as a block,” he wrote. “You are not the only client seeing this[,] so time will be of the essence. Save the price discussion for later—at the moment you might want to figure out whether this portfolio suits your objectives.” Later that day, another internal list of “axes” got circulated. “Over the past few weeks[,] we’ve continued to move several CDO and subprime positions,”
Anthony Kim, another mortgage trader, wrote to the senior members of the mortgage group. He then summarized for them that $859.4 million worth of the stuff had sold from Goldman’s inventory. In a separate e-mail, a congratulatory note went around to the group thanking Robert Gaddi, another trader, for doing a “great job” for “moving us out of [$]6mm of our BBB-, Fremont, subprime risk” along with the request to “continue to focus” on the additional long list of bonds. On April 11, another list circulated. “Please continue to focus on the axes below—they remain a high priority for the desk,” the note read. The list came with a further admonition to the sales force: “We are very axed to move” several tranches of the still unsold
Timberwolf CDO. The
message included the news that the desk was having trouble selling the deal at prices anything like what it had been selling the securities for, and that no longer mattered. “We need levels from accounts that will move this risk,” the message said. “We are planning to pay in the context of $20/bond,” meaning Goldman was willing to sell the bonds at a sufficient discount to get them sold.
By April 19, Sparks was geared up to deal. Attaching the list of the securities Goldman was anxious to sell, he wrote to
V. Bunty Bohra, on the structured products syndicate desk, “Why don’t we go one at a time with some ginormous credits—for example, let’s double the current offering of credit for [T]imberwolf” to make it look more attractive to a buyer. Minutes later, Bohra responded, “We have done that with [T]imberwolf already. Don’t want to roll out any other focus axes until we get some traction there but at the same time, don’t want to stop showing the inventory.” Birnbaum recalled how the frustrations with trying to sell the residual inventory at gradual markdowns reverberated through the firm, until it dawned on people that more drastic steps were needed.
The difficulty Goldman faced in April trying to sell its “axes” into a market that no longer wanted to buy what it was trying to sell, came to a head in mid-May. “Sparks and the [mortgage] group are in the process of considering making significant downward adjustments to the marks on their mortgage portfolio, especially CDOs and CDO squared,”
Craig Broderick, Goldman’s global head of risk management, wrote to his team in his famous May 11 e-mail. “This will potentially have a big P&L impact on us, but also to our clients due to the marks and associated margin calls on repos, derivatives, and other products. We need to survey our clients and take a shot at determining the most vulnerable clients, knock-on implications, etc. This is getting lots of 30th floor attention right now.” Recalled Birnbaum: “So we marked the positions. And started telling the world.”