Friday, December 24, 2010

Confessions of an Econometrician

I regularly harp on the problems of econometric models. There have been financial collapses all over Wall Street because of model failures. Long Term Capital Management and the sub-prime disaster are only two of the most prominent failures of financial models.  Thus, it is good to see that  Emanuel Derman, a former physicist-turned banker, who designed one of the first interest rate models, (He spent 17 years at Goldman Sachs) sounds like Hayek or Mises in his criticism of models. He even has a forthcoming  book about it, Metaphors, Models & Theories. But, it seems he does go a bit too far. FT's Gillian Tett tells us:

"there are no genuine theories in finance because finance is concerned with value, an even more subjective concept than heat or pressure,” Derman observes. And unlike physics, financial experiments are not repeatable – history changes how markets work.
True, there are no constants in the world of finance or economics, of the kind that exist in physics, and thus empirical financial experiments can't be conducted, but this doesn't mean theories can't be built. It sounds like Derman needs to read Human Action to understand that you can indeed have theories, impressive, complex, elaborate theories, without the use of empirical methods. If Derman, however, were to couch his thinking by saying, "There are no mathematical theories that can determine the exact value of an item," I would have to agree.

Despite this short-coming in his view on the nature of theory in the sciences of human action, it is good to hear that a top quant is coming out with a book that appears to shout, "The Quants have no clothes!"

There's more from Gillian Tett, here.

1 comment:

  1. As someone who has supported Austrian economics even when I got my economics and statistics degrees, I agree. As one Stats Prof would say, our models we create are fictions, but if we do our homework right, they will be useful fictions.

    I admit I was expecting the recent Hindenburg Omen signal to forecast a drop in the market in September. That didn't happen, but the indicator could be further refined. I afterwards noticed that the market if measured against the silver or gold, then it did have a noticeable fall during the HO's warning period.