How could sophisticated global finance firms get so crazy and take on so much risk? I'd point to two reasons. First, their math wizards came up with models that told them it was not that risky. In a special report about some of those financial nerds, known as "quants," who built these mathematical models underlying all these mortgage-backed securities, Newsweek (June 8, 2009) recalled Warren Buffett's dictum: "Beware of geeks bearing formulas." The magazine then went on to tell the story of über-geek David X. Li, "who, while working at JPMorgan, created the Gaussian copula function, a formula for determining the correlation between the default rates of different securities." In theory, if one mortgage-backed security defaulted, the model gave bankers an estimate of how many others would default as well. "The apparent genius of the Gaussian copula is its abstraction," said Newsweek. "Rather than relying on the immense amount of data used to figure the odds that a [subprime bundle of mortgages and collateralized debt obligations] might default, Li appeared to have discovered a law of correlation. That is, you didn't need the data; the correlation was just there. Armed with it, quants could price these much faster, and traders could buy and sell them at record speeds. Gaussian was rocket fuel for the CDO market. The global volume of CDO deals went from $157 billion in 2004 to $520 billion in 2006. As more banks got in on the game, the once-large profit margins started to shrink. In order for banks to make the same kind of returns, they had to pack more and more loans into a CDO, essentially making bigger bombs." Needless to say, Li's benign predictions about the correlation between defaults quickly proved wrong in the subprime crisis, when subprime mortgages and their bonds toppled together like dominoes. As Newsweek put it: "Li was on his way to a Nobel Prize when the world blew up."
The second reason they so underpriced the risk was simpler. It is something that happens in every bubble: people who are apparently smart turn out to be really dumb—in large numbers. They buy into a notion that nothing can go wrong. In this case the notion was that housing prices in America would never go down again. As Michael Lewis pointed out in a piece on AIGFP in Vanity Fair (August 2009), the AIG unit was originally used by Wall Street investment banks to insure piles of loans to IBM and GE. Then, in the early 2000s, AIGFP started insuring "messier piles": securities backed by credit card debt, student loans, auto loans, and prime mortgages—anything that generated a cash flow. As Lewis noted, these loans were of such a diverse nature and to so many different parties that the usual risk logic applied: they couldn't all go bust at once. Originally, there were very few subprime mortgages in these piles. But that changed toward the end of 2004. "From June 2004 until June 2007," wrote Lewis, "Wall Street underwrote $1.6 trillion of new subprime-mortgage loans and another $1.2 trillion of so-called Alt-A loans." This expansion was made possible, in part, because AIGFP was ready to insure many of these piles of loans—and made billions of dollars doing it.