The Big Short: Inside the Doomsday Machine

by Michael Lewis

  They never actually finished weighing the soul of Greg Lippmann. Rather, they were interrupted by two pieces of urgent news. The first came in May 2006: Standard & Poor's announced its plans to change the model used to rate subprime mortgage bonds. The model would change July 1, 2006, the announcement said, but all the subprime bonds issued before that date would be rated by the old, presumably less rigorous, model. Instantly, the creation of subprime bonds shot up dramatically. "They were stuffing the channel," said Vinny. "Getting as much shit out so that it could be rated by the old model." The fear of new and better ratings suggested that even the big Wall Street firms knew that the bonds they'd been creating had been overrated.

  The other piece of news concerned home prices. Eisman spoke often to a housing market analyst at Credit Suisse named Ivy Zelman. The simple measure of sanity in housing prices, Zelman argued, was the ratio of median home price to income. Historically, in the United States, it ran around 3:1; by late 2004, it had risen nationally, to 4:1. "All these people were saying it was nearly as high in some other countries," says Zelman. "But the problem wasn't just that it was four to one. In Los Angeles it was ten to one and in Miami, eight-point-five to one. And then you coupled that with the buyers. They weren't real buyers. They were speculators."* The number of For Sale signs began rising in mid-2005 and never stopped. In the summer of 2006, the Case-Shiller index of house prices peaked, and house prices across the country began to fall. For the entire year they would fall, nationally, by 2 percent.

  Either piece of news--rising ratings standards or falling house prices--should have disrupted the subprime bond market and caused the price of insuring the bonds to rise. Instead, the price of insuring the bonds fell. Insurance on the crappiest triple-B tranche of a subprime mortgage bond now cost less than 2 percent a year. "We finally just did a trade with Lippmann," says Eisman. "Then we tried to figure out what we'd done."

  The minute they'd done their first trade, they joined Greg Lippmann's long and growing e-mail list. Right up until the collapse, Lippmann would pepper them with agitprop about the housing market, and his own ideas of which subprime mortgage bonds his customers should bet against. "Any time Lippmann would offer us paper, Vinny and I would look at each other and say no," said Danny Moses. They'd take Lippmann's advice, but only up to a point. They still hadn't gotten around to trusting anyone inside a Wall Street bond department; anyway, it was their job, not Lippmann's, to evaluate the individual bonds.

  Michael Burry focused, abstractly, on the structure of the loans, and bet on pools with high concentrations of the types that he believed were designed to fail. Eisman and his partners focused concretely on the people doing the borrowing and the lending. The subprime market tapped a segment of the American public that did not typically have anything to do with Wall Street: the tranche between the fifth and the twenty-ninth percentile in their credit ratings. That is, the lenders were making loans to people who were less creditworthy than 71 percent of the population. Which of these poor Americans were likely to jump which way with their finances? How much did their home prices need to fall for their loans to blow up? Which mortgage originators were the most corrupt? Which Wall Street firms were creating the most dishonest mortgage bonds? What kind of people, in which parts of the country, exhibited the highest degree of financial irresponsibility? The default rate in Georgia was five times higher than that in Florida, even though the two states had the same unemployment rate. Why? Indiana had a 25 percent default rate; California, only 5 percent, even though Californians were, on the face of it, far less fiscally responsible. Why? Vinny and Danny flew down to Miami, where they wandered around empty neighborhoods built with subprime loans, and saw with their own eyes how bad things were. "They'd call me and say, 'Oh my God, this is a calamity here,'" recalls Eisman.

  In short, they performed the sort of nitty-gritty credit analysis on the mortgage loans that should have been done before the loans were made in the first place. Then they went hunting for crooks and fools. "The first time I realized how bad it was," said Eisman, "was when I said to Lippmann, 'Send me a list of the 2006 deals with high no-doc loans." Eisman, predisposed to suspect fraud in the market, wanted to bet against Americans who had been lent money without having been required to show evidence of income or employment. "I figured Lippmann was going to send me deals that had twenty percent no docs. He sent us a list and none of them had less than fifty percent."

  They called Wall Street trading desks and asked for menus of subprime mortgage bonds, so they might find the most rotten ones and buy the smartest insurance. The juiciest shorts--the bonds ultimately backed by the mortgages most likely to default--had several characteristics. First, the underlying loans were heavily concentrated in what Wall Street people were now calling the sand states: California, Florida, Nevada, and Arizona. House prices in the sand states had risen fastest during the boom and so would likely crash fastest in a bust--and when they did, those low California default rates would soar. Second, the loans would have been made by the more dubious mortgage lenders. Long Beach Savings, wholly owned by Washington Mutual, was a prime example of financial incontinence. Long Beach Savings had been the first to embrace the originate and sell model and now was moving money out the door to new home buyers as fast as it could, few questions asked. Third, the pools would have a higher than average number of low-doc or no-doc loans--that is, loans more likely to be fraudulent. Long Beach Savings, it appeared to Eisman and his partners, specialized in asking homeowners with bad credit and no proof of income to accept floating-rate mortgages. No money down, interest payments deferred upon request. The housing blogs of southern California teemed with stories of financial abuses made possible by these so-called thirty-year payment option ARMs, or adjustable-rate mortgages. In Bakersfield, California, a Mexican strawberry picker with an income of $14,000 and no English was lent every penny he needed to buy a house for $724,000.

  The more they examined the individual bonds, the more they came to see patterns in the loans that could be exploited for profit. The new taste for lending huge sums of money to poor immigrants, for instance. One day Eisman's housekeeper, a South American woman, came to him and told him that she was planning to buy a townhouse in Queens. "The price was absurd, and they were giving her a no money down option adjustable-rate mortgage," says Eisman, who talked her into taking out a conventional fixed-rate mortgage. Next, the baby nurse he'd hired back in 2003 to take care of his new twin daughters phoned him. "She was this lovely woman from Jamaica," he says. "She says she and her sister own six townhouses in Queens. I said, 'Corinne, how did that happen?'" It happened because after they bought the first one, and its value rose, the lenders came and suggested they refinance and take out $250,000--which they used to buy another. Then the price of that one rose, too, and they repeated the experiment. "By the time they were done they owned five of them, the market was falling, and they couldn't make any of the payments."

  The sudden ability of his baby nurse to obtain loans was no accident: Like pretty much everything else that was happening between subprime mortgage borrowers and lenders, it followed from the defects of the models used to evaluate subprime mortgage bonds by the two major rating agencies, Moody's and Standard & Poor's.

  The big Wall Street firms--Bear Stearns, Lehman Brothers, Goldman Sachs, Citigroup, and others--had the same goal as any manufacturing business: to pay as little as possible for raw material (home loans) and charge as much as possible for their end product (mortgage bonds). The price of the end product was driven by the ratings assigned to it by the models used by Moody's and S&P. The inner workings of these models were, officially, a secret: Moody's and S&P claimed they were impossible to game. But everyone on Wall Street knew that the people who ran the models were ripe for exploitation. "Guys who can't get a job on Wall Street get a job at Moody's," as one Goldman Sachs trader-turned-hedge fund manager put it. Inside the rating agency there was another hierarchy, even less flattering to the subprime mortgage bond raters. "At the
ratings agencies the corporate credit people are the least bad," says a quant who engineered mortgage bonds for Morgan Stanley. "Next are the prime mortgage people. Then you have the asset-backed people, who are basically like brain-dead."* Wall Street bond trading desks, staffed by people making seven figures a year, set out to coax from the brain-dead guys making high five figures the highest possible ratings for the worst possible loans. They performed the task with Ivy League thoroughness and efficiency. They quickly figured out, for instance, that the people at Moody's and S&P didn't actually evaluate the individual home loans, or so much as look at them. All they and their models saw, and evaluated, were the general characteristics of loan pools.

  Their handling of FICO scores was one example. FICO scores--so called because they were invented, in the 1950s, by a company called the Fair Isaac Corporation--purported to measure the creditworthiness of individual borrowers. The highest possible FICO score was 850; the lowest was 300; the U.S. median was 723. FICO scores were simplistic. They didn't account for a borrower's income, for instance. They could also be rigged. A would-be borrower could raise his FICO score by taking out a credit card loan and immediately paying it back. But never mind: The problem with FICO scores was overshadowed by the way they were misused by the rating agencies. Moody's and S&P asked the loan packagers not for a list of the FICO scores of all the borrowers but for the average FICO score of the pool. To meet the rating agencies' standards--to maximize the percentage of triple-A-rated bonds created from any given pool of loans--the average FICO score of the borrowers in the pool needed to be around 615. There was more than one way to arrive at that average number. And therein lay a huge opportunity. A pool of loans composed of borrowers all of whom had a FICO score of 615 was far less likely to suffer huge losses than a pool of loans composed of borrowers half of whom had FICO scores of 550 and half of whom had FICO scores of 680. A person with a FICO score of 550 was virtually certain to default and should never have been lent money in the first place. But the hole in the rating agencies' models enabled the loan to be made, as long as a borrower with a FICO score of 680 could be found to offset the deadbeat, and keep the average at 615.

  Where to find the borrowers with high FICO scores? Here the Wall Street bond trading desks exploited another blind spot in the rating agencies' models. Apparently the agencies didn't grasp the difference between a "thin-file" FICO score and a "thick-file" FICO score. A thin-file FICO score implied, as it sounds, a short credit history. The file was thin because the borrower hadn't done much borrowing. Immigrants who had never failed to repay a debt, because they had never been given a loan, often had surprisingly high thin-file FICO scores. Thus a Jamaican baby nurse or Mexican strawberry picker with an income of $14,000 looking to borrow three-quarters of a million dollars, when filtered through the models at Moody's and S&P, became suddenly more useful, from a credit-rigging point of view. They might actually improve the perceived quality of the pool of loans and increase the percentage that could be declared triple-A. The Mexican harvested strawberries; Wall Street harvested his FICO score.

  The models used by the rating agencies were riddled with these sorts of opportunities. The trick was finding them before others did--finding, for example, that both Moody's and S&P favored floating-rate mortgages with low teaser rates over fixed-rate ones. Or that they didn't care if a loan had been made in a booming real estate market or a quiet one. Or that they were seemingly oblivious to the fraud implicit in no-doc loans. Or that they were blind to the presence of "silent seconds"--second mortgages that left the homeowner with no equity in his home and thus no financial incentive not to hand the keys to the bank and walk away from it. Every time some smart Wall Street mortgage bond packager discovered another example of the rating agencies' idiocy or neglect, he had himself an edge in the marketplace: Crappier pools of loans were cheaper to buy than less crappy pools. Barbell-shaped loan pools, with lots of very low and very high FICO scores in them, were a bargain compared to pools clustered around the 615 average--at least until the rest of Wall Street caught on to the hole in the brains of the rating agencies and bid up their prices. Before that happened, the Wall Street firm enjoyed a perverse monopoly. They'd phone up an originator and say, "Don't tell anybody, but if you bring me a pool of loans teeming with high thin-file FICO scores I'll pay you more for it than anyone else." The more egregious the rating agencies' mistakes, the bigger the opportunity for the Wall Street trading desks.

  In the late summer of 2006 Eisman and his partners knew none of this. All they knew was that Wall Street investment banks apparently employed people to do nothing but game the rating agencies' models. In a rational market, the bonds backed by pools of weaker loans would have been priced lower than the bonds backed by stronger loans. Subprime mortgage bonds all were priced by the ratings bestowed on them by Moody's. The triple-A tranches all traded at one price, the triple-B tranches all traded at another, even though there were important differences from one triple-B tranche to another. As the bonds were all priced off the Moody's rating, the most overpriced bonds were the bonds that had been most ineptly rated. And the bonds that had been most ineptly rated were the bonds that Wall Street firms had tricked the rating agencies into rating most ineptly. "I cannot fucking believe this is allowed," said Eisman. "I must have said that one thousand times."

  Eisman didn't know exactly how the rating agencies had been gamed. He had to learn. Thus began his team's months-long quest to find the most overrated bonds in a market composed of overrated bonds. A month or so into it, after they bought their first credit default swaps on subprime mortgage bonds from Lippmann, Vincent Daniel and Danny Moses flew to Orlando for what amounted to a subprime mortgage bond conference. It had an opaque title--ABS East--but it was, in effect, a trade show for a narrow industry: the guys who originated subprime mortgages, the Wall Street firms that packaged and sold subprime mortgages, fund managers who invested in nothing but subprime mortgage-backed bonds, the agencies that rated subprime mortgage bonds, the lawyers who did whatever the lawyers did. Daniel and Moses thought they were paying a courtesy call on a cottage industry, but the cottage was a castle. "There were so many people being fed by this industry," said Daniel. "That's when we realized that the fixed income departments of the brokerage firms were built on this."

  That's also when they made their first face-to-face contact with the rating agencies. Greg Lippmann's people set it up for them, on the condition they not mention that they were betting against, and not for, subprime mortgage bonds. "Our whole purpose," said Moses, "was supposed to be, 'We're here to buy these securities.' People were supposed to think, 'Oh, they're looking to buy paper because it's getting to attractive levels.'" In a little room inside the Orlando Ritz-Carlton hotel, they met with both Moody's and S&P. Vinny and Danny already suspected that the subprime market had subcontracted its credit analysis to people who weren't even doing the credit analysis. Nothing they learned that day allayed their suspicion. The S&P people were cagey, but the woman from Moody's was surprisingly frank. She told them, for instance, that even though she was responsible for evaluating subprime mortgage bonds, she wasn't allowed by her bosses simply to downgrade the ones she thought deserved to be downgraded. She submitted a list of the bonds she wished to downgrade to her superiors and received back a list of what she was permitted to downgrade. "She said she'd submit a list of a hundred bonds and get back a list with twenty-five bonds on it, with no explanation of why," said Danny.

  Vinny, the analyst, asked most of the questions, but Danny attended with growing interest. "Vinny has a tell," said Moses. "When he gets excited he puts his hand over his mouth and leans his elbow on the table and says, 'Let me ask you a question about this...' When I saw the hand to face I knew Vinny was on to something."

  Here's what I don't understand, said Vinny, hand on chin. You have two bonds that seem identical. How is one of them triple-A and the other not?

  I'm not the one who makes those decisions, said the woman from Moody's, but sh
e was clearly uneasy.

  Here's another thing I don't understand, said Vinny. How could you rate any portion of a bond made up exclusively of subprime mortgages triple-A?

  That's a very good question.

  Bingo.

  "She was great," said Moses. "Because she didn't know what we were up to."

  They called Eisman from Orlando and said, However corrupt you think this industry is, it's worse. "Orlando wasn't even the varsity conference," said Daniel. "Orlando was the JV conference. The varsity met in Vegas. We told Steve, 'You have to go to Vegas. Just to see this.'" They really thought that they had a secret. Through the summer and early fall of 2006, they behaved as if they had stumbled upon a fantastic treasure map, albeit with a few hazy directions. Eisman was now arriving home at night in a better mood than his wife had seen him in a very long time. "I was happy," says Valerie. "I thought, 'Thank God there's a place to put all this enthusiastic misery.' He'd say, 'I found this thing. It's a gold mine. And nobody else knows about it.'"

  CHAPTER FIVE

  Accidental Capitalists

  The thing Eisman had found was indeed a gold mine, but it wasn't true that no one knew about it. By the fall of 2006 Greg Lippmann had made his case to maybe 250 big investors privately, and to hundreds more at Deutsche Bank sales conferences or on Deutsche Bank conference calls. By the end of 2006, according to the PerTrac Hedge Fund Database Study, there were 13,675 hedge funds reporting results, and thousands of other types of institutional investors allowed to invest in credit default swaps. Lippmann's pitch, in one form or another, reached many of them. Yet only one hundred or so dabbled in the new market for credit default swaps on subprime mortgage bonds. Most bought this insurance on subprime mortgages not as an outright bet against them but as a hedge against their implicit bet on them--their portfolios of U.S. real estate-related stocks or bonds. A smaller group used credit default swaps to make what often turned out to be spectacularly disastrous gambles on the relative value of subprime mortgage bonds--buying one subprime mortgage bond while simultaneously selling another. They would bet, for instance, that bonds with large numbers of loans made in California would underperform bonds with very little of California in them. Or that the upper triple-A-rated floor of some subprime mortgage bond would outperform the lower, triple-B-rated, floor. Or that bonds issued by Lehman Brothers or Goldman Sachs (both notorious for packaging America's worst home loans) would underperform bonds packaged by J.P. Morgan or Wells Fargo (which actually seemed to care a bit about which loans it packaged into bonds).