The Long and the Short of It

The first smell of blood finally hit the subprime market in November 2006. Dan Sparks first detected it when Kevin Gasvoda phoned him that month with some strange news. The obscure subprime originators, who doled out money advanced by Wall Street dealers in return for home loans that fed the giant securitization pipelines, were starting to wobble.

The incidents recounted by Gasvoda stemmed from a standard protection mechanism Wall Street firms had for their mortgage loan pipelines, called representations and warranties. As part of their contractual agreement to purchase loans from origination companies, firms like Goldman could demand their money back if the loans were obviously defective. It was just like consumers' having the right to return a defective washing machine. Up to that point, Goldman and other dealers had only performed the minimum due diligence on the subprime loans they bought, so confident were they that they could immediately be sold. But now the frauds that had been building up in the subprime mortgage machine were finally coming to the surface. The originators and brokers had scraped the barrel: they were now lending to such hopeless prospects that mortgages would default within a month or two after closing. These early payment defaults obviously couldn't be securitized, which was why Gasvoda was trying to recover the money Goldman had advanced to the mortgage finance companies. “It's a breach of an agreement,” said Sparks. “They have to buy them back.”

As far as Goldman's bottom line was concerned, the issue was little more than an irritant. Most of the subprime loans in its warehouse seemed to be performing and could still be securitized and sold to investors. Indeed, the whole point of the warehouse was to insulate the assets inside it from the ups and downs of the market. It was through the innovation of the subprime derivative broadcasting system that the defaults caused Goldman trouble — but ultimately would save the firm. Returning to the analogy of a layered mortgage bond as a multi-story building in a flood, the Intex models based on house prices and foreclosure rates said that the lower floors were still perfectly safe. But if you took the market prices of default swaps and fed them backwards into Intex, those same lower floors were flooded. That fed into the triple-B-minus category of the ABX index that tracked this mezzanine layer of the building. And because Sparks's traders owned a $6 billion long position in this index, a modest decline in the ABX had an immediately noticeable impact at the firm.

 At the start of December, Goldman's mortgage unit had ten days of consecutive trading losses. The losses weren't large, but they were distressing for a firm not accustomed to losing money.

Closing its 2006 financial year at the end of November, Goldman reported record annual profits, including $14.2 billion from fixed income trading, of which over $1 billion in gross revenue came from mortgages. But was that business now unraveling? Sparks knew how his bosses — president Gary Cohn, CFO David Viniar, and CEO Lloyd Blankfein — would react to the losses. He had just been given a long-awaited promotion to the firm's risk committee, which meant he would now be required to attend the Wednesday morning risk committee meetings that Viniar chaired, and Cohn and Blankfein often attended.

Sparks found himself in an unwelcome spotlight beside his peers. “This is really puzzling that you're having these issues,” said Blankfein thoughtfully. “It doesn't make sense,” added Cohn, more brusque and direct than Goldman's brainy CEO. “Every other market we operate in is booming.” American house prices were flat-lining, not declining, according to Goldman's reckoning. The ten days of losses had not breached any VAR limit, and credit spreads were still at record lows. But the derivatives radar system that Sparks was using could not be ignored. “Look, we get it,” Viniar said. “It's OK if you make less money but we have to reduce the risk.” On December 14, 2006, the CFO assembled Sparks and the senior members of his team for a special meeting.

Goldman needed a lot of mortgages flowing through its pipeline to make the kind of money the mortgage department had been making. The firm reported securitizing around $5 or $6 billion of subprime and commercial real estate loans per month throughout 2006. And at the end of the year, there were over $10 billion of subprime loans sitting inside Goldman as the pipeline bifurcated into the mortgage bond and CDO production lines that Sparks ran. Their prices weren't budging, but the fact that these warehouses of cash securities were invisible to the derivatives radar was a troubling sign. Then Viniar focused on what the radar did pick up: Sparks's trading positions. “You're too big. Let's bring this closer to home,” he told Sparks as the meeting ended. “Let's be flatter. Let's be aggressive distributing things.”

As Sparks communicated Viniar's orders to cut risk, latent tensions among the Goldman mortgage team began to emerge. There was Gasvoda and Ostrem, respectively creating mortgage bonds and CDOs for the firm, and they needed to be long in order to do it. Then there were the traders — David Lehman, Michael Swenson, and Josh Birnbaum — who together with Egol collectively functioned as Goldman's equivalent of Greg Lippmann. Out of the three, Birnbaum was intensely self-confident (even by Goldman standards). It was Birnbaum, say former colleagues, who had purchased the long position in the most threatened layer of the ABX index back in the spring, and who had made a lot of money for Goldman on the trade. Even before Viniar's directive Sparks had had to force the headstrong young trader to cut back. “He was the longest guy in the market,” a senior Goldman trader recalls.

Now Birnbaum would have to go the other way and bet against subprime, in order to cancel out those long positions that Ostrem and Gasvoda had. But how would Goldman perform a 180 without revealing its cards? Because dealers could post ABX bid and offer prices on Reuters terminals and trade with one another using brokers, it was easy to figure out who was trading heavily and which way they were positioned. Birnbaum insisted that he could fool the market. “Given how much ABX we purchased through the broker market, the world would think we were long for the foreseeable future,” he told Sparks.

 While selling off his long ABX trade, Birnbaum proposed selling default swaps on individual mortgage bonds to the most uncritical buyers imaginable: the vast stampede of ABS CDO androids robotically snapping up every mortgage bond they could find. Birnbaum could quietly step in, buying protection or, in other words, shorting subprime bonds that Goldman didn't own.

But Pete Ostrem, who had put together over $10 billion of CDOs for Goldman in 2006 and was keen to beat his record in 2007, expressed alarm. He ran a factory almost recognizable to laypeople: it brought in mortgage bonds as raw materials and converted them to CDOs at a profit. These cash mortgage bonds (and Goldman owned $7 billion of them) hadn't gone down in price the way the derivatives had. This meant that the subprime CDOs that Ostrem built looked less attractive to customers than synthetic CDOs constructed purely from derivatives.

 By selling these subprime derivatives to Goldman's CDO competitors, Birnbaum was undermining Ostrem's business.

Tension between Sparks's underlings was not new. Ostrem had also clashed repeatedly with Jonathan Egol, whose Abacus transactions, constructed out of subprime default swaps, also competed with what Ostrem was doing. The Abacus factory didn't have a warehouse — it simply faced the trading desk. Investors like IKB and AIGFP took the long side, selling protection to Egol, who took the short side, accumulating a position that by 2006 had reached $14 billion in size. Ostrem saw that as a threat, and Sparks had long struggled to defuse frictions between his acolytes. Now he had to keep the headstrong Birnbaum in line. The two rivals on the trading floor would have to work together, Sparks explained. Ostrem would retool his CDO factory to start doing Abacus-style deals, providing Birnbaum with an outlet — invisible to the Street — in terms of subprime default swaps that he needed to short.

Add comment

Security code

2018 © Copyright | Finance and Investment