At around one in the morning on 15 September 2008, Lehman Brothers issued a two-paragraph press release. The fourth-largest investment bank in the United States, 158 years old, was filing for Chapter 11. No rescue, no buyer, no arranged marriage. Bankruptcy.
Everyone had assumed it was too big to fail. What broke that morning was not one bank but the belief that authorities would always step in, that the risks were understood, and that the models measuring them worked. The months that followed produced the deepest global downturn since the 1930s.
From a dry goods store to Wall Street
The firm started in 1844 in Montgomery, Alabama, where Henry Lehman, a 23-year-old immigrant from Bavaria, opened a general store. His brothers Emanuel and Mayer joined him, and the business drifted into cotton trading — first accepting cotton as payment, then brokering it.
They moved to New York in 1858 and built a commodities and then a securities house. The decisive shift came in 1906, when Lehman partnered with a then-small firm called Goldman Sachs to bring General Cigar and Sears, Roebuck public. Underwriting turned out to be a better business than trading cotton, and Lehman spent the next century taking American companies to market.
By 2007 the transformation was complete. Lehman reported revenue of about $19 billion and net income of $4.2 billion, employed roughly 28,000 people, held a AA credit rating and traded at $86 a share in February. Dick Fuld, chief executive since 1993 and known as the Gorilla, had steered it through the 1997 Asian crisis, the dot-com collapse and the aftermath of September 2001. The firm looked as though it had already survived everything that could happen to it.
How the fuel was assembled
Lehman did not fail because of one bad quarter. It failed because of a structure that took most of a decade to build, and every piece of it looked rational while house prices were rising.
Cheap money. After the 2001 recession the Federal Reserve cut its policy rate from 6.5% to 1% and held it there into 2004. Banks could borrow at 1% and lend at 5% or more. The constraint was no longer the cost of funding; it was finding borrowers.
A housing boom. American house prices roughly doubled between 2000 and 2006, with far larger gains in California, Nevada and Florida. Out of that came a working assumption that hardened into doctrine: national house prices do not fall. If that is true, a loan secured on a house is a safe loan regardless of the borrower.
Subprime lending. Mortgages were written for borrowers who would not previously have qualified, structured to hide the problem. A teaser rate of 2% or 3% for two years, then a reset to 8% or higher. Little or no income verification — the industry's own term was NINJA loans, no income, no job, no assets. Loans at 100% of the purchase price, sometimes more, with the early payments covering interest only.
The sales pitch showed only the teaser payment. A household that could afford $800 a month was sold a house whose payment would become $2,400 after the reset. The plan, to the extent there was one, was that the house would be worth 30% more by then and could be sold or refinanced. That plan had exactly one assumption in it.
Securitization: how the risk got everywhere
Local lenders did not keep these mortgages. They sold them within weeks to Wall Street, and this is where Lehman sat in the chain.
The process was mechanical. Buy thousands of mortgages, pool them, slice the pool into tranches with different priority on the cash flows, and sell the tranches as bonds. Mortgage-backed securities, and then collateralized debt obligations built from the leftovers of the first round.
The alchemy was in the ratings. Moody's, Standard & Poor's and Fitch awarded AAA to senior tranches of pools made from loans nobody expected to be repaid on the original terms. The justification was diversification: individual defaults were modeled as broadly independent events, so a large enough pool would only ever lose a small fraction.
Defaults were not independent. They shared one cause — house prices — and when prices turned, the loans failed together, across regions, at once. The correlation assumption was the entire structure, and it was wrong in exactly the scenario it was meant to survive.
The rating agencies were paid by the issuers whose products they rated. That conflict was public, disclosed, and priced by nobody.
Leverage of thirty to one
Lehman, like its peers, ran on borrowed money. By the end of 2007 its leverage ratio was around 30 to 1: roughly $25 billion of equity supporting some $600 billion of assets.
The arithmetic is unforgiving. At 30 to 1, a 3.3% fall in asset values wipes out the entire equity. Lehman held hundreds of billions in mortgage-related securities and commercial real estate whose prices were about to move considerably more than 3.3%.
Worse, most of that borrowing was short-term repo funding that had to be rolled over daily. A bank in that position does not fail when it becomes insolvent; it fails when lenders decline to roll, which can happen in a morning and for reasons that are entirely about perception.
Repo 105
In its final year Lehman used an accounting device to make the leverage look smaller than it was. Under a treatment known internally as Repo 105, it moved assets off the balance sheet just before quarter-end by classifying repo transactions as sales rather than financings, then brought them back days later.
The bankruptcy examiner's report found around $50 billion removed this way at the peak. It was legal opinion-shopped in another jurisdiction because no US law firm would sign it off, and it was the balance sheet equivalent of tidying a room by putting everything in a cupboard for the duration of the inspection.
The timeline of the unraveling
The Fed began raising rates in 2004, and by June 2006 the policy rate was 5.25%. Subprime resets, calculated off short-term rates, arrived exactly as house prices stopped rising. Then came the sequence:
- February–April 2007. New Century Financial, one of the largest subprime lenders, reports heavy losses and files for bankruptcy.
- June 2007. Two Bear Stearns hedge funds holding subprime CDOs collapse.
- August 2007. BNP Paribas freezes three funds, saying it can no longer value the assets. Interbank lending seizes; the ECB injects €95 billion in a single day.
- September 2007. Northern Rock suffers the first British bank run in around 150 years.
- March 2008. Bear Stearns loses its funding in 72 hours. JPMorgan buys it in a Fed-brokered deal at $2 a share, later raised to $10, against $170 the previous year.
After Bear Stearns the market had one question: who is next. The answer was the firm with the most mortgage exposure and the thinnest capital, and everyone could see which one that was.
Lehman reported a $2.8 billion loss in the second quarter of 2008 and a $3.9 billion loss in the third. On 7 September the government took Fannie Mae and Freddie Mac into conservatorship. On 9 September Lehman's shares fell 45% in a day. Attempts to sell the firm to Korea Development Bank and then Bank of America went nowhere.
The weekend
On 12 September the heads of Wall Street's major firms were summoned to the New York Fed. Three options were on the table.
A government rescue was refused. Treasury Secretary Henry Paulson said there would be no public money, and the legal authority for a direct bailout of a broker-dealer was, at best, unclear.
Bank of America would buy Lehman only if the government absorbed losses on tens of billions of impaired assets. Paulson declined, and Bank of America went off to buy Merrill Lynch instead — a deal agreed that same weekend, which removed the other likely buyer from the table.
Barclays was the last option and wanted the deal. It was blocked from London: the Financial Services Authority would not approve a British bank assuming Lehman's obligations without a guarantee, and no government would provide one. By Sunday evening there was nothing left to arrange.
Monday, and the fortnight that followed
The filing listed $613 billion of debt against $639 billion of assets, most of them illiquid and marked at prices no buyer would pay. It remains the largest corporate bankruptcy in US history. Around 26,000 employees lost their jobs, many of them holding deferred compensation in stock that was now worth nothing.
The Dow fell 504 points, 4.4%, on the day. That was not the crisis; it was the starting gun.
On 16 September the Fed lent $85 billion to AIG, which had written credit protection on hundreds of billions of mortgage securities and could not post collateral. The final commitment reached $182 billion.
The same day the Reserve Primary Fund, a money market fund holding Lehman commercial paper, broke the buck — its net asset value fell below a dollar. Money market funds were the instrument treated as equivalent to cash by corporate treasurers everywhere. Redemptions followed immediately, and by 18 September the commercial paper market had frozen, meaning ordinary companies could not borrow to make payroll.
On 29 September Congress rejected the $700 billion TARP package and the Dow fell 777 points, its largest point decline at the time. The bill passed four days later.
How far it traveled
Iceland's three largest banks were nationalized within a week and the country's banking system, several times the size of its economy, ceased to exist. The United Kingdom put together a support package of around £500 billion. Germany rescued Hypo Real Estate at a cost above €100 billion. Ireland guaranteed its entire banking system, roughly €440 billion, and the resulting liability nearly bankrupted the state.
Economies with no direct subprime exposure were hit through trade and funding. World trade fell about 12% in 2009. Euro area GDP contracted around 4.5%, and unemployment climbed for years afterwards. In the United States, unemployment went from 4.7% in 2007 to 10% in late 2009, roughly nine million jobs disappeared, and millions of households lost their homes to foreclosure.
The financial crisis then became a sovereign debt crisis in Europe, as governments that had absorbed bank losses found their own borrowing costs questioned — a second act that ran until 2012 and reshaped the euro area's institutions.
What happened to the people
Dick Fuld ran Lehman for fifteen years and owned a great deal of its stock. His paper fortune fell from around a billion dollars to a small fraction of it. He testified before Congress in October 2008, faced years of civil litigation, and was never criminally charged — as was true of essentially every senior executive at every firm involved. He has maintained since that Lehman was solvent and that the government made a mistake.
Henry Paulson had been chief executive of Goldman Sachs before becoming Treasury Secretary. He declined to rescue Lehman on Sunday and committed $85 billion to AIG on Tuesday, a sequence he has spent the years since explaining: AIG had collateral to lend against and Lehman, in his account, did not. The inconsistency was what markets actually reacted to.
The 26,000 employees produced the images that define the day — people leaving the Times Square headquarters carrying boxes, having learned by email. Many had deferred compensation held in Lehman stock, so they lost their job and their savings in the same hour. Almost none of them had anything to do with the mortgage book.
The absence of criminal convictions is the part that shaped the politics of the following decade more than any regulation did.
What actually changed, and what did not
Changed. Capital requirements are substantially higher and leverage at the largest banks is a fraction of what Lehman ran. Annual stress tests model severe scenarios. Resolution planning exists, so that a large bank failing is meant to be an orderly process rather than an improvisation over a weekend. Bail-in rules are designed to put losses on bondholders before taxpayers.
Not changed. The largest US banks are bigger than they were in 2008, not smaller. Rating agencies are still paid by issuers. Leverage migrated rather than disappeared, into hedge funds, private credit and other lightly supervised vehicles. And regulation has already been partially rolled back — parts of the Dodd-Frank Act were loosened in 2018 for mid-sized banks, several of which ran into trouble in 2023 for reasons that would have been familiar in 2008.
Was letting it fail the right call?
The argument for is that rescuing every institution would have destroyed any remaining discipline, and that Lehman was insolvent rather than merely illiquid, which limited what the Fed could legally lend against.
The argument against is arithmetic. Whatever a rescue would have cost, the contagion cost far more, and the inconsistency was the worst part — Bear Stearns rescued, Lehman abandoned, AIG rescued forty-eight hours later. Markets could no longer tell which institutions were protected, so they treated all of them as unprotected and stopped lending to everyone.
Both positions have something to them. Lehman was probably beyond saving by September 2008. Letting it fail without any mechanism to contain what followed is the part that is hard to defend.
What a trader should take from it
Leverage decides how long you survive being wrong. At 30 to 1, a 3% move is terminal. The position size that looks efficient in a calm market is the reason there is no account left when the market stops being calm.
Correlations converge under stress. The models priced independent defaults. Diversification that depends on things staying uncorrelated fails precisely when it is needed, and this is true of a mortgage pool and of a trading book.
Funding risk kills faster than credit risk. Lehman did not run out of assets. It ran out of lenders willing to roll overnight financing. The same logic applies to a margin account: solvency is irrelevant if you cannot meet the call today.
Complexity hides risk rather than removing it. Nobody at the top of Lehman could value its book in a stressed market, which is why no buyer would take it without a guarantee.
The crisis also permanently changed the assets traders watch. It is the reason central bank balance sheets became a market variable, the reason funding spreads are followed as a stress indicator, and part of the reason interest rates stayed near zero for a decade — which produced the next set of imbalances, in a pattern that rate cycles repeat with some reliability.
