Why AIG Wasn't Allowed to Fail

Mar.17.09 | About: American International (AIG)

Justin Fox wonders whether we should have just let AIG fail -- or at least the holdco and the AIGFP subsidiary. They certainly deserved to fail. So why did we bail them out? Because of the systemic fragility of the CDS market, is the answer -- it's basically the same reason why the government stepped in to prevent Bear Stearns from being forced into liquidation. It feared a cascade of counterparty failures which could kill the entire financial system.

Here's the fear: AIG goes bust, and can no longer make good on the promises it made when it said that it would pay out on a CDS contract in the event that a certain credit defaults. Default protection sold by AIG, in other words, becomes worthless. Now let's say you're a CDS desk at, say, JP Morgan. You're buying and selling default protection all the time, and so long as the amount you've bought, on any given credit, is equal to the amount you've sold, you reckon that you have no net exposure.

The minute that AIG fails, everybody else's net position alters substantially, and in a very unpredictable way. The protection that JP Morgan bought from AIG is worthless, while the offsetting protection that JP Morgan sold to some hedge fund remains outstanding. So JP Morgan now has a large position it never wanted.

Now there's a good chance that JP Morgan will have hedged its counterparty risk to AIG -- but that doesn't make the risk go away, it just shunts it elsewhere in the financial system. And the web of connections between the thousands of counterparties in the CDS market is so complex that no one really has a clue who would have ended up holding the multi-billion-dollar bag. All those AIG losses which are currently being borne by the government wouldn't have disappeared if AIG had failed: they would simply have turned up somewhere else in the financial system.

But no one would have had a clue where in the financial system, exactly, those losses would have ultimately come to rest. And given the magnitude of the losses, you can be sure that no one would have wanted to have any kind of dealings with the poor schmucks who ended up on the hook for all those billions of dollars. And since those poor schmucks could be pretty much anybody, no one would do any kind of business with anybody else: you'd get settlement risk run amok. The entire global financial system could grind to a halt overnight, due to the inability of any given institution to persuade any other institution that it was actually solvent.

We don't know for sure that this kind of worst-case scenario would have happened if AIG had been allowed to fail. But we don't know that it wouldn't have happened -- and the US government felt that it simply couldn't take that kind of risk.

What's more, bailing out AIG had the pleasant side-effect of putting the entire global CDS market on a much stronger footing. Remember that CDS, like all derivatives, are a zero-sum game: for every loser, there's an equal and opposite winner. Very few institutions were net sellers of protection; AIG was by far the largest. So what that means is that the rest of the CDS market, ex AIG, is now a net winner to the exact extent that AIG is a loser: a hundred billion dollars or more. Given worries about the fragility of the CDS market and the systemic risks that it posed, bailing out the single largest net seller of protection essentially meant injecting a large amount of government cash into the part of the market that regulators were most worried about. It was quite an elegant solution, in its way: rather than trying to unpick the CDS knot institution by institution, you could just bail them all out at once by backstopping AIG.

Remember that what regulators were most worried about at the time was systemic risk. Whether or not AIG deserved the money was pretty much beside the point: the key thing was that if it didn't get the money, the entire global financial system would be put at risk of collapse. In which light, the cost of the AIG bailout looks positively modest, compared to its benefit.