The Risks of Financial Modeling

Hasschu

Testimony before congressional committee Sept. 10 2009
The Risks of Financial Modeling: VaR and the Economic Meltdown.
Testimony of David Colander
http://us1.institutionalriskanalytics.com/pub/IRAstory.asp?tag=382

Of interest to the educators in our midst.

nesf

Stopped reading after he basically implied that Physics doesn't deal with complex (i.e. non-linear) systems.

gerry87

nesf wrote: »

Stopped reading after he basically implied that Physics doesn't deal with complex (i.e. non-linear) systems.

Maybe you should have read a bit more... cos I don't think he does. Even if he does, it's not his point.

nesf

gerry87 wrote: »

Maybe you should have read a bit more... cos I don't think he does. Even if he does, it's not his point.

I'll read it alright, just later. Honestly, I find people who say "Natural Sciences straightforward, Social Sciences more complex" amusing. Then, I'm in the unusual position of having studied both at high level.

nesf

Right, I've read the piece.

First off, the title of the thread is completely off the mark. Financial modelling from the last 20 years is not what he's complaining about. He's complaining about a specific type of model mostly used in macroeconomic modelling and forecasting. Some of the model types he lists as likeable are staples of financial modelling.

Secondly, he's rehashing a debate that has raged within economics for the best part of 30 years. The Walrasian models he speaks off were most in vogue during the 70s and early 80s but have become less influential since then. The DGSE model he focuses his criticism on is a hybrid model taking some Walrasian elements and mixing them with neo-Keynesian thinking which is very different to Walrasian thinking. And honestly, you could say it's raged for the best part of a century if you care to go back far enough, but it's most certainly been a major factor since the late 20th century.

Finally, the broad thrust of his argument is something I'm very sympathetic to. Economic systems at their heart are complex non-linear systems. You cannot accurately model these using linear models, they'll match the data some of the time but not when we most need them to, for instance during crashes and booms and whatnot. That said, he paints a far more homogeneous picture of economics than that which is real, though I think this was unintentional on his part given his list of preferred modelling techniques. The DGSE model for instance while the dominant model in macroeconomic forecasting, is not the only model used by central banks and such. We're not, thankfully, in the kind of Walrasian hegemony of the 70s. The single biggest issue is a changing of tools from Walrasian style modelling to more complex non-linear work. The type of mathematics used is extremely different and this makes switching from DGSE to other more complex models very tricky at a profession wide level. I'd also like to highlight that he's complaining about mainstream macroeconomists here rather than economists in general!