In one sentence: Michael Lewis explains the 2008 financial crisis through the few investors who read the subprime mortgage documents nobody else read, bet against the bonds built from them, and spent two years watching the market refuse to agree, using their disbelief to take the machine apart.
At a Glance
Author: Michael Lewis
First published: 2010 (W. W. Norton)
Category: Finance / Narrative Non-Fiction
Length: about 266 pages, about 93,000 words (W. W. Norton hardcover)
ISBN-13: 978-0-393-07223-5 (W. W. Norton hardcover)
Summary reading time: about 12 minutes
Book reading time: about 6 hours
Notable adaptations: the 2015 film directed by Adam McKay, which won the Academy Award for Best Adapted Screenplay and was nominated for Best Picture
Content note: the book reproduces trading-floor language, including crude and contemptuous terms for customers and for immigrant borrowers, and one chapter takes its title from a trader’s remark about a borrower who could not speak English. Chapter 1 recounts the death of Steve Eisman’s infant son, and later chapters discuss Michael Burry’s Asperger’s diagnosis and his son’s. Every figure is as of 2010, before the litigation and settlements that followed. Lewis’s sourcing is largely the participants themselves, which the analysis section takes up.
Lewis had written this story once already. Liar’s Poker, published twenty years earlier, was meant as a warning about the bond desks of the 1980s and read instead as a recruiting brochure. Coming back in 2010 he made the choice that shapes everything: rather than narrate the crisis from the top, he told it through the few investors who bet against the machine. His access was extraordinary, taking in Burry’s own emails and letters and anonymous sources inside Morgan Stanley and Deutsche Bank, and bafflement becomes the explanatory device throughout.
Read it if you want the mechanics of structured credit explained properly, or reporting that moves like a thriller without cheating on the numbers. Skip it if you want a policy history of 2008 or a balanced account of culpability.
The Big Idea
The crisis, on Lewis’s account, was not a panic but a reality that had existed for years finally overwhelming a perception everyone was paid to maintain. The loans had always been bad. The question was when the bonds built on them would be marked at what they were worth, and that depended less on evidence than on whether the firms holding the paper could afford to admit it.
What made the delay possible was a chain in which no link had reason to look down. The broker sold the loan, the lender sold the bond, the bank sold the tranche, and the rating agency was paid per deal by the issuer. The CDO manager collected fees at both ends and did little between. The pension fund could buy only highly rated paper, so it bought the rating rather than the asset.
Which is why the mechanics are the story, and why the ideas below take them in order.
Key Ideas
1. A subprime bond was a tower built to be eaten from the bottom
A subprime mortgage bond pooled thousands of home loans and sold claims on the cash they produced in layers, called tranches. The top layer was paid first and rated triple-A. The bottom absorbed the first losses and was rated triple-B. Losses ate the tower from the ground up. On Lewis’s figures the triple-B tranche went to zero once the pool lost roughly 8 percent, far less than most buyers assumed it could absorb. That floor was thin: about a billion dollars of poor-quality loans might yield only 20 million dollars of lowest triple-B tranche, which is why demand for product outran the supply of borrowers.
2. The credit default swap turned a private opinion into a position
Subprime bonds could not be sold short. A credit default swap on one was insurance against its default: the buyer paid a small annual premium, often around 2 percent of the face amount, and collected the full face value if the bond failed. What Lewis stresses is how strange this was to want. It let an outsider bet against a named pool of loans without owning any part of it, the loss capped at the premiums and the gain the whole notional, and it fixed the timing problem that ruins bearish trades. Paperwork was the other wall: Cornwall Capital had to be granted an ISDA master agreement before the big dealers would trade with them.
3. The CDO laundered triple-B into mostly triple-A
The thin bottom layers had to go somewhere. A collateralized debt obligation took the triple-B tranches from about a hundred different mortgage bonds, stacked them into a new tower, and presented the result to the agencies as a diversified portfolio. On Lewis’s account they pronounced roughly 80 percent of that new tower triple-A. His image for the trick is a hundred buildings on the same floodplain, every ground floor equally exposed. When real triple-B tranches ran short, the synthetic CDO arrived, the same tower built from swaps rather than loans, so the shorts’ own purchases supplied the raw material.
4. The models rated averages, and the loans were designed to reset later
Two failures did most of the work. The agencies’ models leaned on FICO scores, and crucially on the average score of a pool, so a pool half-filled with very good credit and half with very bad looked identical to one of uniform mediocrity. Pools were assembled to hit the average that maximized the triple-A slice. Loan structures compounded it. Teaser rates, interest-only periods and adjustable resets meant the borrower’s real ability to pay was tested about two years after the bond was rated, long after the paper was sold. Lewis reports that the agencies held pool-level data only and assumed house prices would keep rising.
5. Nobody along the chain was paid to look at the loans
Burry’s edge was comically low-technology. A physician turned investor running Scion Capital from Cupertino, he read the loan prospectuses themselves, which Lewis says almost nobody else did, and watched the share of interest-only adjustable loans climb while the reported averages held steady. Over his first five years Scion was up 242 percent against an S&P 500 down 6.84 percent. Eisman and his FrontPoint team came the other way, from equity research and a career covering subprime lenders. Greg Lippmann of Deutsche Bank assembled the other side and hunted for clients, which is where the bet became available at scale.
6. Howie Hubler is proof that seeing it was not enough
The book’s smartest structural decision is its dark mirror. Howie Hubler at Morgan Stanley worked out that the triple-B tranches were bad and bought protection on them, the crown jewel a 2 billion dollar bespoke position he expected to return 2 billion in profit. The running premiums were his problem, and one of his own traders puts it plainly: the desk was supposed to earn 2 billion dollars a year, and the swap position was costing 200 million. Hubler covered that by selling protection on triple-A CDO tranches at 28 basis points, eventually about 16 billion dollars of them, which he treated as very nearly risk-free. They were built from the same triple-B bonds. The agencies had assumed roughly 30 percent correlation between the underlying pools, and the real figure was closer to total. Risk reports filed the position as triple-A and read it as safe. On the first 4 billion dollars the loss came to roughly 3.7 billion. Lewis writes that Hubler lost more money than any single trader in Wall Street history and kept what he had already been paid.
7. The collapse was a fight over marks, not a revelation
When the loans finally failed, the prices did not move, because the dealers set the marks. Burry’s swaps were valued against his counterparties’ own books, his fund bled premiums through 2006 and 2007, and he side-pocketed it as investors demanded their money back. Lippmann had to demand 1.2 billion dollars of collateral from Morgan Stanley on 4 billion dollars of swaps, pricing the same CDOs at 70 when Morgan Stanley’s model said 95. Cornwall Capital, begun with 110,000 dollars in a Schwab account and a shed in Berkeley, sold out from a Devon pub in August 2007, netting more than 80 million dollars and growing from a bit over 30 million dollars to 135 million. The turn came when Meredith Whitney, an obscure analyst at Oppenheimer, argued on 31 October 2007 that Citigroup had mismanaged itself into needing a dividend cut. Lewis’s framing is that she was not calling the bankers corrupt. She was calling them stupid.
Context and Analysis
What holds up is most of it. This is the best popular explanation of structured credit anyone has written, and it works because Lewis explains through a character’s incomprehension rather than a schematic. The reporting got access nobody else had and the detail has largely survived scrutiny. The central argument, that the machine ran on incentives rewarding everyone for not looking, has been confirmed rather than overturned by what came after. For the theory behind the failure, go to The Black Swan: the agencies priced an instrument whose entire risk sat in the tail using a short and benign stretch of housing history, Taleb’s argument made concrete at a scale of trillions.
The frame does real distorting work, and the clearest problem is survivorship. We meet the people who were right and could hold the position. The investors who reached the same conclusion a year too early and were redeemed out of existence are absent, and they are the evidence that would say whether the scarce thing was the insight or the funding. The heroes also get an easy ride. Eisman’s earlier career involved the subprime lenders he later attacked, and the shorts’ purchases of swaps were the material synthetic CDOs were assembled from, so the bet against the machine helped feed it. Lewis notices this and files it as irony. His sourcing is the participants, with Burry’s story resting on his own emails and the Hubler chapter on anonymous sources, and where versions conflict Lewis settles it in a footnote, finding Hubler’s account of a dispute with a superior less plausible.
Other weaknesses deserve naming. The CDO as presented is stylized, a pure pile of triple-B tranches with 80 percent rated triple-A, where real deals were messier, and Lewis’s footnotes concede that banks besides his characters supplied much of the other side of AIG’s exposure. The villains are caricatured where analysis would be harder, the agencies staffed by the second-rate and the German buyers merely credulous. The epilogue traces 2008 to a single cause, Gutfreund’s 1981 decision to take Salomon Brothers public and move risk from partners to shareholders. Figures drift too: Burry’s premium is 200 basis points in one place and 250 in another, his profit both 720 million and 750 million dollars. The book stops in 2010, before the settlements. As a study of institutional blindness it pairs with Bad Blood, John Carreyrou’s Theranos account, where sophisticated people deferred to one another’s confidence and an outsider reconstructed the truth from documents anyone could have read.
What It Teaches
A few things to take from it:
1. Read the primary documents. The prospectuses existed, and almost nobody opened them. 2. Averages hide the distribution that matters, and a model fed an average will misprice a barbell. 3. Ask what the people closest to a market are paid to believe before trusting their confidence. 4. Being right early is indistinguishable from being wrong unless you can fund the position. 5. Correlation assumptions are where catastrophic risk hides, since one shock reaches every address on the floodplain.
Memorable Lines
“Why bother to overturn your parents’ world when you can buy it and sell off the pieces?” (Lewis)
“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.” (Lewis)
“Here was a strange but true fact: The closer you were to the market, the harder it was to perceive its folly.” (Lewis)
“If the oranges were rotten, the orange juice was also rotten.” (Lewis)
“He was smart enough to be cynical about his market but not smart enough to realize how cynical he needed to be.” (Lewis)
“The problem wasn’t that Lehman Brothers had been allowed to fail. The problem was that Lehman Brothers had been allowed to succeed.” (Lewis)
Should You Read the Full Book?
Verdict: Essential
A summary can carry the mechanics, and the mechanics are the argument, so most of it survives compression. What does not survive is what makes them stick. The book delivers them through accumulating disbelief, in the voices of people whose money depended on being right, and the January 2007 dinner in Las Vegas with a CDO manager explains the market’s blindness better than any summary can. Read it for the reporting, the dialogue and the sensation of a trillion-dollar error coming into focus. The Big Short book page has the full details and where to get a copy.