AIG Didn't Lose the Bet. It Got Margin-Called to Death.

On September 15, 2008, AIG did not default on anything. No bond it insured had failed. No mortgage pool had crossed the line where its promises came due. What happened that day was smaller and far more lethal: three rating agencies cut AIG below AA-, and a clause buried in thousands of derivative contracts woke up. Within hours, counterparties were demanding $32 billion in collateral, and AIG was short $12.4 billion of the cash to deliver it.⁶ The largest insurer on earth was not insolvent. It was illiquid — margin-called to death by a downgrade, not a default.

The official story is that AIG made a reckless half-trillion-dollar bet on subprime mortgages and lost. Almost every part of that sentence is wrong. The headline number — $527 billion, the figure that made AIG a synonym for greed — is the notional value of its credit default swap book as of year-end 2007: the face amount of the contracts, not the loss.¹² AIG never lost anything close to that. What it lost was control of its own cash flow.

The notional number is the magic trick

Start with the scale, because the scale is where the myth lives. At its peak, AIG's super-senior CDS portfolio carried gross notional exposure of $441 billion — but roughly $379 billion of that (the corporate-loan and prime-mortgage tranches) was written for financial institutions, principally in Europe, purely for regulatory capital relief, contracts that never had any meaningful subprime risk in them at all.⁹¹⁰ Strip those away and the dangerous book — the multi-sector CDOs stuffed with mortgage bonds — was about $72 billion notional, of which roughly $61 billion touched subprime.⁷ Notional is the amount of stuff being insured. It is not, and never was, the amount at risk. Saying AIG 'lost $500 billion' is like saying a fire insurer 'lost' the full appraised value of every house in town the moment one chimney smoked.

So how much did the toxic book actually cost? AIG's combined unrealized CDS losses for 2007 and 2008 were estimated at $33.2 billion.⁷ A serious wound — but $33 billion is a tenth of the number the public remembers, and 'unrealized' is doing heavy lifting. These were mark-to-market losses on contracts that, in many cases, would never be called. The man who built the book, Joseph Cassano, told the FCIC plainly: 'I did not expect actual, economic losses on the portfolio.'⁸ He was largely right about the underwriting. He was catastrophically wrong about the plumbing.

How a downgrade does what a default couldn't

Here is the mechanism the popular story skips, and it is the whole story. A credit default swap is a promise: AIG agreed to pay out if the insured bonds went bad. But the contracts had a second, quieter feature — a collateral schedule. If the market value of the insured securities fell, or if AIG's own credit rating dropped, AIG had to post cash to its counterparties as security against a future payout that might never come. Two triggers, both independent of any actual default. As subprime CDOs were marked down through 2007 and 2008, the first trigger started firing. When the rating agencies cut AIG below AA- on September 15, the second trigger fired all at once. The promise hadn't come due. The deposit on the promise had.⁶

Why the rescue was a profit and not a loss

If you understand the crisis as a liquidity event rather than an underwriting failure, the bailout's strange ending stops being strange. The government wasn't buying garbage; it was lending against assets that were temporarily unmarkable but not actually worthless. The first move, on September 22, 2008, was an $85 billion revolving credit facility from the New York Fed — priced at a punitive LIBOR-plus-8.5%, the terms of a lender who expected to be paid back.² When that proved too blunt, the November 10 restructuring reshaped the whole thing: the credit line shrank to $60 billion, and the Fed stood up Maiden Lane III to simply buy the troubled CDOs outright — with AIG taking the first $5 billion of any loss.³ The state didn't absorb the toxic book. It warehoused it and waited.

The waiting paid. Across the Fed and Treasury, the total commitment reached roughly $182.3 billion — $112.5 billion from the Fed, $69.8 billion from Treasury via TARP.⁴ But 'committed' is not 'spent,' and 'spent' is not 'lost.' As the markets thawed, the warehoused CDOs recovered most of their value, AIG sold assets and repaid, and Treasury unwound its equity stake by December 2012. The final tally: a combined positive return of $22.7 billion — $17.7 billion to the Fed, $5.0 billion to Treasury.⁴ The most notorious bailout in history made money. It could only have done that if the thing being rescued was never as broken as the headline number claimed.

Wasn't AIG just reckless, and Cassano the villain?

The honest objection is that 'liquidity, not solvency' lets AIG off too easily. And there's truth in it — the book really was recklessly built. AIGFP's super-senior CDS exposure rocketed from $17.9 billion at the end of 2004 to $54.3 billion a year later, much of the expansion crammed into a single calendar year, and the tranches were packed with subprime, Alt-A, and payment-option ARM mortgages.⁵ Cassano told the FCIC he stopped writing subprime swaps in late 2005, but the documents show AIGFP still wrote a $433.5 million CDS in January 2006 on a CDO whose collateral pool was 93% subprime.⁵ This was not prudence. It was a firm selling earthquake insurance and forgetting it had to keep cash on hand for the aftershocks.

But notice the part the villain narrative quietly drops. The CDS book wasn't even AIG's only liquidity hole. A separately run securities-lending program — nothing to do with Cassano — bled roughly $21 billion in 2008 by lending out bonds and reinvesting the cash collateral into the same souring mortgage assets.⁸ Two unrelated parts of AIG ran the same hidden bet: that they'd never need a lot of cash all at once. Both were wrong in the same month. The lesson isn't 'one rogue desk blew up a giant.' It's that an entire firm had quietly become a maturity-mismatch machine, and the mismatch — not the mortgages — is what detonated.

AIG is remembered as the company that bet the house on subprime and lost. The truer epitaph is stranger and more useful: it was a company that was roughly right about the bet, dead wrong about the cash, and saved by a government that figured out the difference and turned a profit on the gap. The bonds it insured mostly didn't default. The promise mostly held. What broke was the firm's ability to keep its footing while the market demanded it prove, in cash, today, that it could honor a promise it would probably never have to keep. The fall wasn't about being wrong. It was about being unable to wait to be proven right.