An old subject, but I had missed this fine piece of financial journalism.
- In 2000, while working at JPMorgan Chase, Li published a paper in The Journal of Fixed Income titled "On Default Correlation: A Copula Function Approach." (In statistics, a copula is used to couple the behavior of two or more variables.) Using some relatively simple math—by Wall Street standards, anyway—Li came up with an ingenious way to model default correlation without even looking at historical default data. Instead, he used market data about the prices of instruments known as credit default swaps.
False. They had calculated the probability of it happening to be a "six sigma event" or six orders of magnitude outside the standard deviation of probability. A two in a billion chance. It's in here somewhere, I think. Pretty sure Lehman actually ran the numbers. Somewhere there's a hoary old list of "you might be an engineer if" that includes the line "you've ever modeled a horse as a sphere." Garbage In, Garbage Out - if your assumptions are crap, your model is crap. If your model is crap, your predictions are crap. It's worth noting that Credit Default Swaps were created by junior execs at Goldman Sachs while they were plastered on margaritas at a weekend retreat in Cabo San Lucas.The reason that ratings agencies and investors felt so safe with the triple-A tranches was that they believed there was no way hundreds of homeowners would all default on their loans at the same time. One person might lose his job, another might fall ill. But those are individual calamities that don't affect the mortgage pool much as a whole: Everybody else is still making their payments on time.
Li's breakthrough was that instead of waiting to assemble enough historical data about actual defaults, which are rare in the real world, he used historical prices from the CDS market. It's hard to build a historical model to predict Alice's or Britney's behavior, but anybody could see whether the price of credit default swaps on Britney tended to move in the same direction as that on Alice.
I totally agree. In fact, I read the spirit of your quoted section as they did a bunch of fancy technological analysis and fooled themselves into believing what they already wanted to believe: that there was no risk and all upside. It reminds me of an exchange of my other piece of financial journalism on the crisis, This American Life episode "The Giant Pool of Money": It's easy to ignore your gut fear when you're making a fortune in commissions. But Mike had other help in rationalizing what he was doing, technological help. Mike sat at a desk with six computer screens connected to millions of dollars worth of fancy analytic software designed by brilliant Ivy League graduates hired by his firm. And this software analyzed all the loans in all the pools that Mike bought and then sold. And the software, the data, didn't seem worried at all. All the data that we had to review, to look at, on loans that were in production, that were years old, was positive. They performed very well. All those factors, when you look at all the pieces and parts, and you say, well, a 90% no income loan three years ago is performing amazingly well. It has a little bit of risk. Instead of defaulting 1 and 1/2% of the time, it defaults 3 and 1/2% of the time. Well, that's not so bad. If I'm an investor buying that, if I get a little bit of additional return, I'm fine. Adam Davidson Wait, Alex, I want step in here, because this is a very important piece of tape. A big part of this whole story, the whole crisis, is that a lot of really smart people, people who knew better, fooled themselves with this data. It was the triumph of data over common sense. Can you play that tape again? Alex Blumberg Yeah, sure. Here you go. All the data that we had to review, to look at, on loans that were in production, that were years old, was positive. Adam Davidson As we now know, they were using the wrong data. They looked at the recent history of mortgages and saw that the foreclosure rate is generally below 2%. So they figured absolute worst-case scenario, the foreclosure rate might go to 8% or 10% or even 12%. But the problem with that is that there were all these new kinds of mortgages given out to people who never would have gotten them before. So the historical data was irrelevant. Some mortgage pools today are expected to go beyond 50% foreclosure rates.Alex Blumberg
Mike Francis Mike Francis
Might have been the WSJ - can't find it now - that mentioned stock statistics look fuckin' gangbusters right now because the Great Recession is no longer reflected in their 10-year returns. https://www.cnbc.com/2019/03/04/the-10th-anniversary-of-the-climactic-march-2009-market-bottom-arrives-this-week.html And since everything has had spectacular performance (once you disregard the 40% decline), everything is a buy!