Not many of my readers work in derivatives, I suspect, but a lot of you work in computer programming, and I think that this blog post from The Economist may resonate with you. The blogger is talking about how “quants” (people with top skills in esoteric math such as theoretical physicists) devised various complex financial instruments, but were back office staff and were not responsible for how they were used. That job fell to the sales suits in the front office, who didn’t have much idea how these things actually worked. Sound familiar? So now you have a good idea why we are in this mess.
As some of the commenters point out, many of the instruments were not that complex in and of themselves. It was what people did with them that was the problem. In software we have learned by painful experience that new releases need to be tested, often in extreme conditions. Too many banks appear to have had the equivalent of poor testing procedures that were more geared to getting the product out of the door than finding out whether it was bug free.
I’ve worked for those kinds of companies — but we weren’t testing banking or medical software… sheesh!
It isn’t so much testing the software as testing the results of the modeling. You have a model that tells you how much profit or loss you will make under a variety of market conditions. That calculation requires a lot of assumptions, and a bunch of algorithms. The algorithms you can test, the assumptions not so much. And in some cases the results of the tests are so scary that the pointy-hairs put their hands over their ears and refuse to listen.