Without Dodd-Frank, AIG Could Happen Again

The near-failure of insurance firm American International Group Inc. (AIG) in 2008 was arguably the most perilous moment of the global financial crisis. Unfortunately, something similar could easily happen again.

The Dodd-Frank Wall Street Reform and Consumer Protection Act contains ample measures aimed at preventing a repeat of the AIG debacle. It gives regulators new powers to oversee insurance companies and other nonbank financial institutions. It sheds more light on the credit derivatives that got AIG in trouble, so regulators can see if any company is taking on too much risk. It requires that parties to derivatives contracts put up adequate collateral -- cash that would cover debts if one or more of them went bust.

Problem is, Dodd-Frank ignores the crucial element that ultimately brought AIG to the brink and threatened to topple its counterparties: the way demands for collateral can come all at once, particularly when market turmoil turns seemingly safe bets into big risks.

Consider what happened to AIG. The company had sold a form of insurance, known as credit-default swaps, on pools of securities -- largely subprime mortgage bonds -- with a total face value of about $55 billion. The deals, struck in 2004 and 2005, required AIG to make big payments if losses to defaults in the bond pools exceeded a certain threshold, often more than 30 percent of the face value of the underlying bonds. The threshold was so high, and AIG’s credit so strong, that the banks buying the insurance -- including Societe Generale SA, Goldman Sachs Group Inc. and Deutsche Bank AG -- initially demanded very little collateral.

Loss Position

By 2007, the housing market was in free fall, and defaults were accelerating on a trajectory that could breach AIG’s threshold. As a result, AIG’s swaps were in a loss position: Even if the insurance payments weren’t triggered, the company would have to pay to get out of the contracts. At this point, AIG’s counterparties should have demanded a lot more collateral to make sure the company would make good on its obligations.

Some tried. Goldman Sachs, for example, demanded $1.8 billion in July 2007. AIG disputed the call on the grounds that Goldman had improperly calculated its losses -- an easy claim to make because the swaps were rarely traded and hence lacked an objective market value. For such an illiquid position, AIG’s and Goldman’s valuation models could provide different answers.

Others didn’t ask for more collateral, either because they maintained a lazy faith in AIG’s creditworthiness or because they had significant mortgage bets of their own and didn’t want to roil markets by exposing AIG’s predicament.

The counterparties’ collateral deficiency left them in a tough spot in 2008, when further deterioration in markets hit AIG’s credit rating and sharply increased the company’s unrealized losses. In a matter of a few months, AIG’s swap counterparties demanded roughly $21 billion in added payments, more than in the previous 13 months combined.

The sudden rush turned the collateral calls into a crisis for everyone involved. AIG didn’t have enough cash on hand to make the payments, a situation tantamount to bankruptcy. If AIG failed, its counterparties -- including several systemically important banks -- would suffer billions of dollars in losses on the unpaid collateral. In some cases, those losses could be big enough to trigger more failures. Only the government bailout of AIG prevented a disaster.

Collateral Calls

A couple of modest tweaks could prevent such a situation from arising in the future.

First, regulators should impose a limit on the amount of collateral that can be demanded on illiquid positions. The limit, for example, could be 10 percent of the notional value of the contract in any two-week period. This constraint would force complacent or conflicted parties to be more vigilant because they would no longer be able to wait until the eleventh hour to request collateral. It would also help prevent collateral calls from overwhelming a company’s available cash, as happened with AIG.

Second, regulators must provide counterparties with an impartial mechanism to resolve valuation disputes. The dispute between Goldman and AIG demonstrates that Dodd-Frank cannot -- as it does -- rely on financial modeling to define adequate collateral levels.

Here’s how it could work. Goldman would submit its disputed collateral request to a regulator (possibly the Securities and Exchange Commission) and immediately cancel its swap contract with AIG. Within 24 hours, Goldman would purchase a new, identical contract to establish an objective market price. Whichever side’s valuation proved wrong would then have to pay the other side twice the difference.

The mechanism could help the entire market by providing a benchmark for hard-to-value contracts. The potential penalty might also help avert disputes by giving counterparties an incentive to be honest with their valuations in the first place.

These changes alone, of course, would not have prevented AIG’s troubles, which were the result of poor risk controls, negligent oversight by regulators, bad assessments by ratings companies and reckless internal management. But together with the other measures in Dodd-Frank, they could make the system much more resilient next time around.

(Matthew Schoenfeld is a third-year student in the JD program at Harvard Law School. Prior to Harvard, he was a member of the macro trading team at 3G Capital, a New York-based hedge fund. The opinions expressed are his own.)

To contact the writer of this article:

To contact the editor responsible for this article: Mark Whitehouse at