How AIG Failed

The WSJ shines a bit more light on what went wrong at AIG today, with a story centering on the chap who designed its risk models, Gary Gorton. In a nutshell, anybody writing credit protection runs two risks: the default risk of the underlying security, on the one hand, and market risk, on the other. It seems that AIG only ever asked Gorton to worry about default risk; no one ever bothered to calculate the risk that CDS spreads would gap out, forcing AIG to take billions of dollars in mark-to-market losses and post many billions of dollars more in collateral.

Mr. Gorton's models harnessed mounds of historical data to focus on the likelihood of default, and his work may indeed prove accurate on that front. But as AIG was aware, his models didn't attempt to measure the risk of future collateral calls or write-downs, which have devastated AIG's finances...
AIG began selling credit-default swaps around 1998. Mr. Gorton's work "helped convince Cassano that these things were only gold, that if anybody paid you to take on these risks, it was free money" because AIG would never have to make payments to cover actual defaults, according to the former senior executive at the unit. However, Mr. Gorton's work didn't address the potential write-downs or collateral payments to trading partners.

Incidentally, AIG Financial Product's Joseph Cassano was kept on as a consultant after he retired in February -- at a rate of $1 million per month.

The WSJ also reports, without citing even anonymous sources, on the subject of Goldman's exposure to AIG, and the amount of AIG collateral that Goldman now holds:

Goldman Sachs Group Inc., for instance, has pried from AIG $8 billion to $9 billion, covering virtually all its exposure to AIG -- most of it before the U.S. stepped in.

Goldman protected itself in other ways, too, since it's not easy getting that collateral:

Late last October, Goldman asked for even more collateral, $3 billion. Again, AIG disagreed, and it ultimately posted $1.5 billion. Goldman hedged its exposure by making a bearish bet on AIG, buying credit-default swaps on AIG's own debt, according to one person knowledgeable about this move...
Mr. Gorton attended the Federal Reserve Bank of Kansas City's annual gathering in Jackson Hole, Wyo. He presented a 92-page paper, "The Panic of 2007," which explained how the financial markets came unglued after a series of unexpected events, such as when clients of financial firms suddenly sought to reclaim assets put up as collateral. "It is difficult to convey," he wrote, "the ferocity of the fights over collateral."

At heart, here, is an age-old debate over the value of any fixed-income instrument. Let's say you buy a bond at par which makes all its interest and principal payments in full and on time. Then you're happy, and making money. But let's say that a couple of years after issue, that bond is trading at just 10 cents on the dollar. Have you lost money?

As far as AIG was concerned, it was one of the biggest companies in the world, more than capable of weathering any mark-to-market storm -- and therefore all it cared about was default risk, not market risk. But as a result, it took on much more market risk than it was really aware of -- and that market risk ended up forcing the entire company into the arms of the US government.

The lesson, of course, is simple, but hard to learn: it's not the risks you measure which bring you down, it's the risks you don't measure. But protecting against those risks is very, very hard.