On September 15, 2008, Lehman Brothers, a firm that had survived 158 years of wars, panics, and depressions, filed for the largest bankruptcy in American history. Over $600 billion in assets were suddenly frozen, and within days the global financial system came within hours of complete collapse. The story most people tell about this catastrophe is simple: greedy bankers handed out loans to people who could not pay them back, Wall Street packaged those loans into products nobody understood, and the whole world paid the price. That story is not false.

But it mistakes the smoke for the fire. The real crisis was not a morality tale about individual greed. It was the failure of an entire system, an invisible architecture of debt, mathematics, deregulation, and global money flows that had been under construction for nearly two decades. Subprime mortgages were the spark, but not the cause.

Even at the worst of the crisis, total losses on those loans amounted to a few hundred billion dollars, a serious sum, but nowhere near enough to explain why the entire global economy nearly seized up. To understand why a wound of that size proved almost fatal, you have to understand the patient’s underlying condition. That means going back to something almost nobody outside finance had heard of before 2008: the shadow banking system. Traditional banking works on deposits.

You put money in a bank, the bank lends it out, and the government insures your deposit to prevent the panics that caused the Great Depression. By the early 2000s, a second banking system had grown up alongside this one, largely outside the reach of those protections. Instead of deposits, it ran on repurchase agreements, better known as repo. Institutional investors, money market funds, pension funds, corporations with cash to park overnight, would lend enormous sums to banks and investment firms for short periods, sometimes just a single night, in exchange for collateral, usually bonds.

Trillions of dollars moved through this market every day, and almost nobody outside the industry knew it existed. The economist Gary Gorton, who has spent much of his career studying financial panics, argued that what happened in 2007 and 2008 was not fundamentally different from the bank runs of the nineteenth century. It was a run, just not the kind anyone was watching for. Instead of ordinary depositors lining up outside a bank, it was institutional lenders in the repo market quietly refusing to renew their short-term loans, night after night, to banks whose collateral they no longer fully trusted.

There were no newsreel images of panicked crowds because the people running were sophisticated fund managers moving money with a few keystrokes. But the underlying mechanism, a sudden collective loss of confidence in the safety of short-term debt, was the same one that had toppled banks for two centuries. So what was the collateral backing all this borrowing? This is where mortgages enter the picture, but not in the way the simple story tells it.

Beginning in the 1980s and accelerating through the 2000s, banks stopped holding the loans they issued. Instead, they bundled thousands of individual mortgages into pools and sliced those pools into bonds sold to investors worldwide. This process, securitization, spread risk and freed up capital in theory. But it also did something more dangerous: it severed the connection between the person making the loan and the person bearing the consequences if it went bad.

A mortgage broker earning a commission on volume had every incentive to write as many loans as possible and no financial stake in whether the borrower could repay, because the loan would be sold off within weeks. The discipline of a local banker who knew he would absorb the loss simply evaporated. Wall Street went further. Investment banks took the riskier slices of those mortgage pools and repackaged them into structures called collateralized debt obligations.

Then they created synthetic CDOs, instruments that contained no real mortgages at all, just side bets in the form of credit default swaps, insurance-like contracts wagering on whether other mortgage bonds would default. Investors could now build an essentially unlimited number of these synthetic products, all referencing the same underlying pool of home loans, multiplying total financial exposure to the housing market far beyond the value of the houses themselves. All of these complex instruments needed to be priced and rated so investors around the world could understand how risky they were. The tool that came to dominate this process was a piece of mathematics called the Gaussian copula, developed around the year 2000 by a quantitative analyst named David X.

Li. It solved a problem that had stumped Wall Street for years: how to estimate the odds that many different mortgages held by many different people in many different places might default at the same time. The formula allowed banks to compress an almost impossibly complicated question about human behavior into a single number. But it rested on a catastrophic assumption.

Reliable data on national mortgage defaults only stretched back a decade or two, and that period had been one of almost uninterrupted rising home prices. The model implicitly assumed that a nationwide collapse of housing values was essentially impossible. When home prices did fall everywhere at once, the model broke completely, and it took the ratings built on top of it down with it. The credit rating agencies, Moody’s, Standard & Poor’s, and Fitch, were supposed to answer one question for investors: how safe is this bond?

For decades their ratings had been broadly reliable. But there was a structural conflict of interest at the heart of their business model. They were paid not by the investors who relied on their judgments, but by the banks whose securities they were rating. As the volume of mortgage securities exploded, the agencies competed fiercely for these enormous fees.

Internal documents later revealed by congressional investigators showed agency employees privately joking that a particular deal could be structured by cows and still get a top rating, while executives worried openly about losing market share to a more generous rival. Between 2000 and 2007, the agencies stamped the vast majority of subprime mortgage bonds with investment-grade ratings, and a striking share received AAA, the same seal of safety given to the bonds of the United States government itself. When the housing market turned, Moody’s alone would downgrade the overwhelming majority of the mortgage securities it had rated AAA just a couple of years earlier. In far too many cases, the ratings had not measured risk.

They had laundered it. None of this could have grown to planet-threatening scale without one more ingredient: deregulation. In the late 1990s, Brooksley Born, who chaired a relatively obscure agency called the Commodity Futures Trading Commission, began raising alarms about a rapidly growing, almost entirely unregulated market in over-the-counter derivatives, including the credit default swaps that would later feature so heavily in the crisis. She proposed simply studying whether this opaque market needed oversight.

The response from Federal Reserve Chairman Alan Greenspan, Treasury Secretary Robert Rubin, and his successor Lawrence Summers was swift and unified. These sophisticated instruments traded between sophisticated parties, they argued, needed no government supervision, and raising the question could destabilize the markets. Born was politically isolated and resigned in 1999. The following year, a bill called the Commodity Futures Modernization Act was passed, tucked, as a rider, inside an 11,000-page government spending package with almost no public debate.

It explicitly barred federal regulators from touching the market for credit default swaps and similar instruments. President Clinton signed it into law in December 2000, almost too fittingly, the day after the Supreme Court decision that settled the presidential election, when the nation’s attention was fixed entirely elsewhere. Credit default swaps are in essence insurance contracts. Ordinary insurance is tightly regulated, and companies must hold capital in reserve to make sure they can pay claims.

Because of the 2000 act, credit default swaps faced no such requirement. Firms could sell unlimited quantities of this insurance-like protection without setting aside a single dollar. One company did exactly that on a staggering scale. The insurance giant AIG built up a portfolio of credit default swaps insuring hundreds of billions of dollars of mortgage securities, collecting premiums for years while setting aside almost nothing.

Much of this business had been sold to European banks, particularly in Germany and France, which used the swaps not for protection but as a regulatory trick to make their own balance sheets look safer than they were. When housing cracked, this single American company’s promises became a hidden thread tying the fate of banks across Europe to the health of subprime borrowers in California and Nevada. One more element completed the picture. In 2005, a Federal Reserve economist named Ben Bernanke, soon to become chairman, introduced an idea called the global savings glut.

Countries like China, along with other developing Asian economies and the oil exporters of the Middle East, had accumulated enormous savings, partly as a defensive reaction to the Asian currency crisis of the late 1990s. Rather than investing at home, they funneled hundreds of billions of dollars into what looked like the safest investment on Earth, United States government debt. This flood of foreign capital pushed interest rates down and kept them down. Investors desperate for returns better than government bonds increasingly turned to the mortgage-backed securities Wall Street was manufacturing at record speed.

Part of what inflated the American housing bubble was not simply American greed. It was a genuine worldwide surplus of capital searching the globe for a home and finding one, disastrously, in the American mortgage market. At home, after the dot-com bubble burst in 2000 and the September 11 attacks rattled the economy, the Federal Reserve under Greenspan cut interest rates eleven times in succession, holding the benchmark rate near a historic low of one percent for roughly a year. Cheap money searched for yield, and it poured into housing.

Greenspan himself encouraged home buyers to consider adjustable-rate mortgages, whose payments could reset sharply higher once the initial low period ended. When home prices began rising in every region of the country simultaneously, many economists, including Greenspan, saw it not as a bubble but as a durable shift. Starting in 2006, home prices began falling almost everywhere at once. Adjustable-rate mortgages reset to payments homeowners could not afford.

Delinquencies climbed. The first serious tremors hit in the summer of 2007 when two hedge funds run by the investment bank Bear Stearns collapsed. In March 2008, Bear Stearns itself teetered on the edge of failure as its repo lenders simply refused to keep rolling over its short-term loans, a textbook bank run conducted between institutions. The Federal Reserve engineered an emergency sale to JPMorgan Chase, backstopping billions in risky assets with public money.

Whether policymakers intended it or not, this set an expectation that when a systemically important firm faced mortal danger, the government would step in. That expectation shattered in September 2008. Lehman Brothers found itself in the same position Bear Stearns had faced six months earlier, unable to fund itself, unable to find a buyer after a potential deal with Barclays collapsed at the last moment. Over a tense weekend at the Federal Reserve Bank of New York, the government did not step in.

Treasury Secretary Henry Paulson and Fed officials later gave differing explanations, some pointing to a lack of legal authority, others acknowledging a deliberate desire to draw a line. Whatever the reason, on September 15, Lehman filed for the largest bankruptcy in American history. The very next day, the Federal Reserve extended an emergency loan of $85 billion, eventually ballooning past $180 billion, to rescue AIG. On the same day Lehman failed, the Reserve Primary Fund, a money market fund holding some of Lehman’s now worthless short-term debt, saw its value fall below the sacred one dollar per share threshold.

In the industry, this is remembered simply as breaking the buck. It triggered a genuine run, with over a hundred billion dollars pulled from similar funds within days. The repo market seized up almost completely. Banks stopped trusting banks.

The entire global plumbing of short-term credit came within what several participants later described as a matter of days, possibly hours, of complete failure. Congress eventually passed the Troubled Asset Relief Program, authorizing up to $700 billion to stabilize the financial system. Combined with the Federal Reserve’s emergency lending and its balance sheet expansion through quantitative easing, the total scale of government support has been estimated at well over $10 trillion. Much of the money was repaid, and some programs actually turned a profit.

But the human cost was devastating. Unemployment climbed past ten percent. Roughly nine million jobs disappeared. Nearly six million families lost their homes to foreclosure.

Household wealth fell by more than $10 trillion. And because the American mortgage market had been sold to investors worldwide, European banks were dragged into a crisis that would set the stage, within a couple of years, for a separate sovereign debt crisis across the eurozone. The honest history holds two truths together. Systems do not build themselves.

People built this one, staffed it, chose to leave derivatives unregulated, chose to keep profits flowing as their own risk models flashed warnings, chose market share over honesty. And yet the crisis was not simply the work of a handful of villains. It was the story of an invisible financial architecture built quietly over two decades, by regulators who declined to regulate, mathematicians who mistook elegance for accuracy, rating agencies that mistook fees for integrity, and a global economy so interconnected that a wave of mortgage defaults in American suburbs could threaten the solvency of banks in Frankfurt within weeks.

The crisis that looked like it arrived overnight in the autumn of 2008 had in truth been under construction, one deregulatory choice and one mispriced risk at a time, for the better part of twenty years.