The Smartest Guys in the Housing Market

Despite his reservations, Sparks was already applying the new innovations to add millions to Goldman's profits. Röthig was so pleased with his tailor-made synthetic CDO that he demanded more, and Sparks and Egol set up a production line to crank out the new products, which would be given the Abacus brand name. Egol was the mortgage equivalent of what people in the world of corporate default swaps call a correlation trader.

Correlation trading is all about selling just a few slices of a CDO to investors and replicating the rest by doing proprietary trading in the underlying default swaps, with the help of a model such as the Gaussian copula. Sparks was skeptical about using models to extract money from illiquid new markets like subprime derivatives, and he kept a close watch on what Egol was doing.

Meanwhile, in the non-derivatives world, Goldman's subprime mortgage pipeline was starting to deliver the goods, and in recognition of his successes, Sparks was promoted to run mortgage bond origination as well as trading. Gasvoda had built relationships on the ground; during its 2004 financial year, Goldman issued over $30 billion in subprime mortgage bonds.

 On paper at least, Sparks had another job, as chief executive of the Goldman Sachs Mortgage Company; run by Gasvoda and employing over two hundred people, this company bought the underlying loans from the originators across America and repackaged them into dozens of newborn mortgage robots with names like GSAMP Trust. Sparks was one of the people who signed the SEC registration documents. But that was just a sideline from his main job, running Goldman Sachs's U.S. mortgage business.

Meanwhile, Pete Ostrem was busy creating CDOs for Goldman asset manager clients. He was very different from Egol, the geeky derivatives wizard. While he would use some of the new subprime derivatives to manufacture CDOs, Ostrem preferred buying the underlying “cash” bonds whenever possible. It was what Sparks approvingly called an “originate, clean up, and move on” business. After taking its fee — which was about 1 percent of the proceeds and was often taken in the form of an equity stake at the lowest, riskiest layer of the CDO — Goldman could discreetly step out of the picture and prepare for its next deal.

There was one obstacle. Ostrem could keep the CDO managers happy with the help of subprime derivatives to “deliver the longs” that they demanded. But there was one part of the CDO that the managers weren't interested in, because the returns were too paltry to attract investors. The super-senior tranche, sitting above the triple-A piece that the manager did find investors for, was Goldman's challenge. It was a problem because Ostrem would have bought the $1 billion worth of subprime mortgages that went into the CDO and held them on Goldman's balance sheet using repo borrowing. But unless he could sell that topmost slice of the CDO, Goldman would still have to “fund” several hundred million dollars' worth of assets, something it was loath to do.

This was also the problem with the Abacus CDOs Egol was pumping out. The models that ratings agencies used might have said the risk of losing money was as close to zero as the laws of probability would allow, but Sparks, who was responsible for the risk of Goldman's mortgage portfolio, still wasn't convinced — he wanted to shed that risk.

Once again, help would come from London, the territory of Michael Sherwood. With his salesforce, Sherwood not only would hold the key to shifting Goldman's hard-to-shift product, such as the topmost layers of Sparks's CDOs, but also had figured out a way to act as middleman for the rest of Wall Street. The trick was pulled off with a careful reading of the Basel banking rules. European banks, unlike those in America, did not have absolute leverage ratios restricting the size of their balance sheets. The only thing holding them back was the risk weighting, or “speed limit,” applied to different types of lending. However, if European banks could buy triple-A-rated assets and combine them with a default swap transacted with a bank as counterparty, their capital ratio could be as small as 1.6 percent. That amounted to a leverage ratio of more than 60 to 1, or even higher, if you took into account the softness of European rules on what qualified as capital.

Although super-senior pieces of ABS CDOs did not provide enough of a return for normal investors, for highly leveraged European banks the paltry returns could be sufficiently magnified to become an attractive proposition, even after paying a default swap premium to a counterparty. It was different from the way LB Kiel invested with UBS, where at least the Germans had a stated interest in getting exposure to the U.S. real estate market. Here, armed with a triple-A rating and the protection of a default swap, an investor had little incentive to investigate what they were buying. The larger European banks, such as Société Générale or UBS, were already buying triple-A-rated CDOs to cash in on this regulatory arbitrage and were asking dealers to sell them default swaps on their investments. The Goldman European sales force had a list of second-tier banking clients on the continent, firms like Germany's cooperative savings bank Deutsche Zentral-Genossenschaftsbank, Switzerland's Zürcher Kantonalbank, and Dutch agricultural lender Rabobank, and wanted to sell them an entire package of CDOs and default swaps. But where was the all-important default swap protection going to come from?

Sherwood was on good terms with Joseph Cassano, the chief executive of AIG Financial Products (AIGFP), who lived in London. Founded in the 1980s and backed by its parent's triple-A rating, AIGFP had developed a reputation for making huge, complex derivatives trades linked to interest rates, equities, or currencies that even Goldman wasn't big enough to handle. These were famously lucrative trades from which Cassano and his team kept 30 percent of the profits. The problem of hedging super-senior risk wasn't new, especially for those dealers who constructed CDOs out of riskier components. However, in Cassano, Goldman had a man who had been involved in the market from the beginning, right back to when J.P. Morgan pitched him its groundbreaking BISTRO deal in 1998, a man so confident in selling this unusual form of protection that he would sell Goldman as much super-senior subprime CDO hedges as it wanted — eventually totaling $23 billion.

It was a quintessential Goldman move. Sherwood was a specialist in what one might call “elephant hunting.” The deal that he and Addy Loudiadis orchestrated in 2001 to help the Greek government conceal billions of debt using secret swap contracts was one example of his prowess. By channeling business through a single large counterparty, Goldman could steer clear of the credit and reputational risks that came with spreading business across a range of smaller clients, one of whom might default or complain about the transaction.

While Sherwood remained invisible in the background, Goldman salespeople swung into action, led by Andy Davilman, who was based in New York. Goldman would stand in the middle of two back-to-back default swap transactions: buying protection from AIGFP on one side and passing it on to the European banking client on the other. A former Salomon trader working for Goldman in New York, Ram Sundaram, would be the middleman. Sparks, meanwhile, was invited to London, where he met Cassano's underlings and then attended meetings at AIGFP's U.S. headquarters in Wilton, Connecticut. For Sparks, here was a suitable “elephant” that could do two things: provide a default swap sweetener to persuade European banks to buy the super-senior pieces of Ostrem's deals, and provide a direct hedge to the Goldman trading desk on Egol's Abacus deals.

Sparks already knew American International Group, whose American General life insurance subsidiary was a heavy buyer of the mortgage bonds Goldman issued. But this elephant was on a different planet. AIGFP resembled the hedge fund Long-Term Capital Management (LTCM), another secretive financial firm run by a mixture of sophisticated derivatives geeks and aggressive risk takers. There was Gary Gorton, a brilliant finance professor from the Wharton School who had a consulting agreement to build AIGFP's models.

Unlike LTCM, this elephant was backed by the world's biggest, triple-A-rated insurance conglomerate. AIG's regulatory structure would serve as a lesson in how an assortment of unconnected, well-intentioned regulations could be exploited by highly motivated innovators. In 1999, AIG asked the Office of Thrift Supervision to be its holding company supervisor. The OTS was supposed to regulate small savings and loans institutions, but if the law permitted a sleepy regulator to take fees for supervising a giant conglomerate, why should it complain?

This proved very convenient in 2004, when the European Union agreed to recognize certain U.S. regulators as being equivalent to its own financial supervisors. AIG had set up a bank subsidiary in Paris, Banque AIG. The French bank supervisor, the Commission Bancaire, didn't look too closely at Banque AIG because EU law deferred to the OTS as being responsible.

Of course, Banque AIG was really just a front for AIGFP, which shared premises with the bank's office in London. The bank was a shell where Cassano booked his default swap trades. But because it was a European bank, it fell under the Basel umbrella and ensured the capital benefits from the super-senior default swap trades with the likes of Société Générale.

Cassano set some conditions with Goldman. Rather than relying on credit ratings, AIGFP would send Gorton and his other quants to crawl over Goldman's CDO factory, tweaking the machinery if necessary to ensure they were happy with the mortgages underlying each deal before writing protection. For Cassano, having Gorton was like having a one-man ratings agency in his back pocket, and he enjoyed showing him off to the analysts who tracked AIG's stock. So confident was Cassano in his moonlighting professor that, over time, AIGFP would end up selling protection on $500 billion of CDOs. The elephant was truly in the bag.

Goldman insisted on one condition. Triple-A-rated counterparties such as AIG traditionally refused to sully themselves with the two-way collateral agreements between lower-ranked mortals in the derivatives market. Many dealers accepted this status quo. But for all except a few triple-A sovereign clients, Goldman refused to do business without such agreements. Pricing credit at its market value was almost an article of faith at Goldman, and derivatives counterparties were no exception to the rule. Reassured by Gorton's actuarial analysis, Cassano and Andrew Forster, a former J.P. Morgan banker who ran AIGFP's credit trading desk in London, were relaxed about Goldman's request. At least in 2005, the super-senior layer of CDOs barely traded, staying hidden on the balance sheets of banks or insurance companies. What possible event could result in a price decline sufficient to trigger a collateral call? “They thought the market was so stable it didn't seem like a big give,” one of the Goldman team recalls. Agreeing to Goldman's terms would prove a fatal mistake for AIGFP.

The opportunity to buy subprime CDO protection from AIGFP lasted for about eighteen months. At the end of 2005, Gorton's actuarial model told Cassano that it was time to stop selling protection on subprime CDOs, but by the time that last deal went through in the summer of 2006, AIGFP had written protection on seventy different deals, with a total subprime exposure of $80 billion.

Add comment

Security code

2018 © Copyright | Finance and Investment