Tuesday, February 24, 2009

The Key Problem With Quantitative Models

Felix Salmon has a really interesting article about a particular finance model created by a Chinese mathematician by the name of David Li, based on a statistical technique called Gaussian copula formulation, that was applied to the pricing of financial assets.

I don't agree that this model, of similar ones for that matter, was what caused the crisis. The key problem was monetary policy. And I don't think this was the only flawed quantitative model that was used. But it was clearly a flawed quantitive model that caused great losses for financial institutions. It was based, as you can read in the article, on the kind of thinking Dogbert expresses here.

Another key problem with the model, which is also a more general problem with these kinds of quantitative models, is that it assume that past correlations will always hold true in the future and that you can therefore use past correlations to predict future events. This is not just a problem in finance, but in economics in general, where for example it is assumed that certain forms of fiscal stimulus have fixed multipliers (with economists arguing over how high these multipliers are).

But as Austrian economics teaches us, there are no fixed quantitative correlations in economics (At least not any exact correlations. Relationships based on sound economic theory, such as that between monetary inflation and price inflation, will roughly hold for most of the time, but always to a varying extent and with a varying time lag). A quantitative relationship, even if estimated correctly for one period in time (and there are reasons to doubt that econometricians really does that in most cases), is a unique historical event applicable only to the time and place that the correlation was derived from. It is not a general economic theoretical relationship. The only general economic theoretical relationships are the verbal logic praxeological ones. Li's model, like so many others, assumed that quantitative relationships would always hold, an assumption that many on Wall Street acted on, and which created big losses once the fallacy of that assumption was revealed.


Blogger Ke said...

Someone needs to explain this to Irwin Kellner. He says the worst of the recession is past us because of statistical evidence. He's also the same guy that said the worst of the mortgage crisis was behind us in 2007 and that we should buy bank stocks in 2008.

How does this person have his job? I'd be fired if I were consistently wrong.

6:21 PM  

Post a Comment

<< Home