Credit crunch
In 2004 I did an internship in Deutsche Bank’s Credit Structuring team. It was an incredibly lucrative business, I remember listening in to a phone call with a big European defence agency who bought some structured assets and in the process, made the team of 4 guys I was working with (we were the Germany desk) quite a few million Euro in profit, for what seemed like not a lot of work.
I remember thinking something was weird, how a bunch of bonds that in aggregate didn’t pay much in interest, could be magically combined, and then sliced up, and out of nowhere, 200 or 300 basis points of profit would appear. Surely something to do with the ratings was being manipulated.
Anyway, the profits being made weren’t really profits, in that the ‘profit’ would accrue over the lifetime of the asset, assuming the underlying assets wouldn’t default. But, in terms of bankers’ incentives, no one could care less about how that asset played out as:
1) there was a model showing that the chance of things going tits up was near on impossible
2) bonuses would be in a few months time, and it would be decided by paper profits made that year
Anyhow, the reason I’m blogging this is that I just found out the Deutsche Bank credit traders just lost 100 million Euros, and that several desks have now been shut down. Further, I hear that Lehman Brothers have canned 8% of their global workforce in a month, that Goldman Sachs has a hiring freeze, and shrunk its graduate intake by 25%.
So it’s kind of a big problem. All those structured assets obfuscate
a) what the real level of risk is, and
b) who the hell owns that risk
(Of course I am heavily simplifying).
The Economist has been covering the credit/American sub-prime problem as a leader for the past month or so now. Not that I follow things in much detail, but I’m reminded that in the financial markets uncertainty is the norm, not the exception, and to my final point: you should read The Black Swan.
For further implications of the credit crunch, read how in the UK Northern Rock needed to be bailed out, almost GBP 2bn was withdrawn in the past 3 days, and that investment banks globally are set to lose $30bn.
**
As an aside, the big problem with a lot of financial models is that they assume a normal distribution on variables where they shouldn’t. A normal distribution is fine for independent variables, but in most things where you are modelling human behaviour, things are far from independent (we are heavily influenced by what other people are doing; Facebook Apps are a good example of that). This in turn leads to problems where models heavily underestimate risks of rare events (the normal distribution inherently understates the probability of rare events).
