The Big Short: Inside the Doomsday Machine (11 page)

AIG Financial Products was created in 1987 by refugees from Michael Milken's bond department at Drexel Burnham, led by a trader named Howard Sosin, who claimed to have a better model to trade and value interest rate swaps. Nineteen eighties financial innovation had all sorts of consequences, but one of them was a boom in the number of deals between big financial firms that required them to take each other's credit risks. Interest rate swaps--in which one party swaps a floating rate of interest for another party's fixed rate of interest--was one such innovation. Once upon a time, Chrysler issued a bond through Morgan Stanley, and the only people who wound up with credit risk were the investors who bought the Chrysler bond. Chrysler might sell its bonds and simultaneously enter into a ten-year interest rate swap transaction with Morgan Stanley--and just like that, Chrysler and Morgan Stanley were exposed to each other. If Chrysler went bankrupt, its bondholders obviously lost; depending on the nature of the swap, and the movement of interest rates, Morgan Stanley might lose, too. If Morgan Stanley went bust, Chrysler, along with anyone else who had done interest rate swaps with Morgan Stanley, stood to suffer. Financial risk had been created out of thin air, and it begged to be either honestly accounted for or disguised.

Enter Sosin, with his supposedly new and improved interest rate swap model--even though Drexel Burnham was not at the time a market leader in interest rate swaps. There was a natural role for a blue-chip corporation with the highest credit rating to stand in the middle of swaps and long-term options and the other risk-spawning innovations. The traits required of this corporation were that it not be a bank--and thus subject to bank regulation, and the need to reserve capital against risky assets--and that it be willing and able to bury exotic risks on its balance sheet. It needed to be able to insure $100 billion in subprime mortgage loans, for instance, without having to disclose to anyone what it had done. There was no real reason that company had to be AIG; it could have been any triple-A-rated entity with a huge balance sheet. Berkshire Hathaway, for instance, or General Electric. AIG just got there first.

In a financial system that was rapidly generating complicated risks, AIG FP became a huge swallower of those risks. In the early days it must have seemed as if it was being paid to insure events extremely unlikely to occur, as it was. Its success bred imitators: Zurich Re FP, Swiss Re FP, Credit Suisse FP, Gen Re FP. ("Re" stands for Reinsurance.) All of these places were central to what happened in the last two decades; without them, the new risks being created would have had no place to hide and would have remained in full view of bank regulators. All of these places, when the crisis came, would be washed away by the general nausea felt in the presence of complicated financial risks, but there was a moment when their existence seemed cartographically necessary to the financial world. AIG FP was the model for them all.

The division's first fifteen years were consistently, amazingly profitable--there wasn't the first hint that it might be running risks that would cause it to lose money, much less cripple its giant parent. In 1993, when Howard Sosin left, he took with him nearly $200 million, his share of what appeared to be a fantastic money machine. In 1998, AIG FP entered the new market for corporate credit default swaps: It sold insurance to banks against the risk of defaults by huge numbers of investment-grade public corporations. The credit default swap had just been invented by bankers at J.P. Morgan, who then went looking for a triple-A-rated company willing to sell them--and found AIG FP.
*
The market began innocently enough, by Wall Street standards.

Large numbers of investment-grade companies in different countries and different industries were indeed unlikely to default on their debt at the same time. The credit default swaps sold by AIG FP that insured pools of such loans proved to be a good business. By 2001, AIG FP, now being run by a fellow named Joe Cassano, could be counted on to generate $300 million a year, or 15 percent of AIG's profits.

But then, in the early 2000s, the financial markets performed this fantastic bait and switch, in two stages. Stage One was to apply a formula that had been dreamed up to cope with corporate credit risk to consumer credit risk. The banks that used AIG FP to insure piles of loans to IBM and GE now came to it to insure much messier piles, which included credit card debt, student loans, auto loans, prime mortgages, aircraft leases, and just about anything else that generated a cash flow. As there were many different sorts of loans, to different sorts of people, the logic that had applied to corporate loans seemed to apply to them, too: They were sufficiently diverse that they were unlikely all to go bad at once.

Stage Two, beginning at the end of 2004, was to replace the student loans and the auto loans and the rest with bigger piles consisting of nothing but U.S. subprime mortgage loans. "The problem," as one AIG FP trader put it, "is that something else came along that we thought was the same thing as what we'd been doing." The "consumer loan" piles that Wall Street firms, led by Goldman Sachs, asked AIG FP to insure went from being 2 percent subprime mortgages to being 95 percent subprime mortgages. In a matter of months, AIG FP, in effect, bought $50 billion in triple-B-rated subprime mortgage bonds by insuring them against default. And yet no one said anything about it--not AIG CEO Martin Sullivan, not the head of AIG FP, Joe Cassano, not the guy in AIG FP's Connecticut office in charge of selling his firm's credit default swap services to the big Wall Street firms, Al Frost. The deals, by all accounts, were simply rubber-stamped inside AIG FP, and then again by AIG brass. Everyone concerned apparently assumed they were being paid insurance premiums to take basically the same sort of risk they had been taking for nearly a decade. They weren't. They were now, in effect, the world's biggest owners of subprime mortgage bonds.

Greg Lippmann
watched his counterparts at Goldman Sachs find and exploit someone else's willingness to sell huge amounts of cheap insurance on subprime mortgage bonds and pretty much instantly guessed the seller's identity. Word spread quickly in the small world of subprime mortgage bond creators and traders: AIG FP was now selling credit default swaps on triple-A-rated subprime bonds for a mere 0.12 percent a year. Twelve basis points! Lippmann didn't know exactly how Goldman Sachs had persuaded AIG FP to provide the same service to the booming market in subprime mortgage loans that it provided to the market for corporate loans. All he knew was that, in rapid succession, Goldman created a bunch of multibillion-dollar deals that transferred to AIG the responsibility for all future losses from $20 billion in triple-B-rated subprime mortgage bonds. It was incredible: In exchange for a few million bucks a year, this insurance company was taking the very real risk that $20 billion would simply go
poof
. The deals with Goldman had gone down in a matter of months and required the efforts of just a few geeks on a Goldman bond trading desk and a Goldman salesman named Andrew Davilman, who, for his services, soon would be promoted to managing director. The Goldman traders had booked profits of somewhere between $1.5 billion and $3 billion--even by bond market standards, a breathtaking sum.

In the process, Goldman Sachs created a security so opaque and complex that it would remain forever misunderstood by investors and rating agencies: the synthetic subprime mortgage bond-backed CDO, or collateralized debt obligation. Like the credit default swap, the CDO had been invented to redistribute the risk of corporate and government bond defaults and was now being rejiggered to disguise the risk of subprime mortgage loans. Its logic was exactly that of the original mortgage bonds. In a mortgage bond, you gathered thousands of loans and, assuming that it was extremely unlikely that they would all go bad together, created a tower of bonds, in which both risk and return diminished as you rose. In a CDO you gathered one hundred different
mortgage bonds
--usually, the riskiest, lower floors of the original tower--and used them to erect an entirely new tower of bonds. The innocent observer might reasonably ask, What's the point of using floors from one tower of debt simply to create another tower of debt? The short answer is, They are too near to the ground. More prone to flooding--the first to take losses--they bear a lower credit rating: triple-B. Triple-B-rated bonds were harder to sell than the triple-A-rated ones, on the safe, upper floors of the building.

The long answer was that there were huge sums of money to be made, if you could somehow get them re-rated as triple-A, thereby lowering their perceived risk, however dishonestly and artificially. This is what Goldman Sachs had cleverly done. Their--soon to be everyone's--nifty solution to the problem of selling the lower floors appears, in retrospect, almost magical. Having gathered 100 ground floors from 100 different subprime mortgage buildings (100 different triple-B-rated bonds), they persuaded the rating agencies that these weren't, as they might appear, all exactly the same things. They were another diversified portfolio of assets! This was absurd. The 100 buildings occupied the same floodplain; in the event of flood, the ground floors of all of them were equally exposed. But never mind: The rating agencies, who were paid fat fees by Goldman Sachs and other Wall Street firms for each deal they rated, pronounced 80 percent of the new tower of debt triple-A.

The CDO was, in effect, a credit laundering service for the residents of Lower Middle Class America. For Wall Street it was a machine that turned lead into gold.

Back in the 1980s, the original stated purpose of the mortgage-backed bond had been to redistribute the risk associated with home mortgage lending. Home mortgage loans could find their way to the bond market investors willing to pay the most for them. The interest rate paid by the homeowner would thus fall. The goal of the innovation, in short, was to make the financial markets more efficient. Now, somehow, the same innovative spirit was being put to the opposite purpose: to hide the risk by complicating it. The market was paying Goldman Sachs bond traders to make the market less efficient. With stagnant wages and booming consumption, the cash-strapped American masses had a virtually unlimited demand for loans but an uncertain ability to repay them. All they had going for them, from the point of view of Wall Street financial engineers, was that their financial fates could be misconstrued as uncorrelated. By assuming that one pile of subprime mortgage loans wasn't exposed to the same forces as another--that a subprime mortgage bond with loans heavily concentrated in Florida wasn't very much like a subprime mortgage bond more concentrated in California--the engineers created the illusion of security. AIG FP accepted the illusion as reality.

The people who worked on the relevant Goldman Sachs mortgage bond trading desk were all extremely intelligent. They'd all done amazingly well in school and had gone to Ivy League universities. But it didn't require any sort of genius to see the fortune to be had from the laundering of triple-B-rated bonds into triple-A-rated bonds. What demanded genius was finding $20 billion in triple-B-rated bonds to launder. In the original tower of loans--the original mortgage bond--only a single, thin floor got rated triple-B. A billion dollars of crappy home loans might yield just $20 million of the crappiest triple-B tranches. Put another way: To create a billion-dollar CDO composed solely of triple-B-rated subprime mortgage bonds, you needed to lend $50 billion in cash to actual human beings. That took time and effort. A credit default swap took neither.

There was more than one way to think about Mike Burry's purchase of a billion dollars in credit default swaps. The first was as a simple, even innocent, insurance contract. Burry made his semiannual premium payments and, in return, received protection against the default of a billion dollars' worth of bonds. He'd either be paid zero, if the triple-B-rated bonds he'd insured proved good, or a billion dollars, if those triple-B-rated bonds went bad. But of course Mike Burry didn't own any triple-B-rated subprime mortgage bonds, or anything like them. He had no property to "insure" it was as if he had bought fire insurance on some slum with a history of burning down. To him, as to Steve Eisman, a credit default swap wasn't insurance at all but an outright speculative bet against the market--and this was the second way to think about it.

There was also a third, even more mind-bending, way to think of this new instrument: as a near-perfect replica of a subprime mortgage bond. The cash flows of Mike Burry's credit default swaps replicated the cash flows of the triple-B-rated subprime mortgage bond that he wagered against. The 2.5 percent a year in premium Mike Burry was paying mimicked the spread over LIBOR
*
that triple-B subprime mortgage bonds paid to an actual investor. The billion dollars whoever had sold Mike Burry his credit default swaps stood to lose, if the bonds went bad, replicated the potential losses of an actual bond owner.

On its surface, the booming market in side bets on subprime mortgage bonds seemed to be the financial equivalent of fantasy football: a benign, if silly, facsimile of investing. Alas, there was a difference between fantasy football and fantasy finance: When a fantasy football player drafts Peyton Manning to be on his team, he doesn't create a second Peyton Manning. When Mike Burry bought a credit default swap based on a Long Beach Savings subprime-backed bond, he enabled Goldman Sachs to create another bond identical to the original in every respect but one: There were no actual home loans or home buyers. Only the gains and losses from the side bet on the bonds were real.

And so, to generate $1 billion in triple-B-rated subprime mortgage bonds, Goldman Sachs did not need to originate $50 billion in home loans. They needed simply to entice Mike Burry, or some other market pessimist, to pick 100 different triple-B bonds and buy $10 million in credit default swaps on each of them. Once they had this package (a "synthetic CDO," it was called, which was the term of art for a CDO composed of nothing but credit default swaps), they'd take it over to Moody's and Standard & Poor's. "The ratings agencies didn't really have their own CDO model," says one former Goldman CDO trader. "The banks would send over their own model to Moody's and say, 'How does this look?'" Somehow, roughly 80 percent of what had been risky triple-B-rated bonds now looked like triple-A-rated bonds. The other 20 percent, bearing lower credit ratings, generally were more difficult to sell, but they could, incredibly, simply be piled up in yet another heap and reprocessed yet again, into more triple-A bonds. The machine that turned 100 percent lead into an ore that was now 80 percent gold and 20 percent lead would accept the residual lead and turn 80 percent of that into gold, too.