May 9, 2005

Financial Risk Models

From the Mises Institute blog:

Although the mechanical philosophy is long dead and buried, our age is not without its own dogma regarding properly scientific explanations. Today, the prevailing belief is that any real science must be composed of mathematical models, models which yield quantitative predictions about some class of events based on particular, initial conditions, also specified numerically. Once again, the currently popular methodology has been imposed on diverse disciplines with little regard to whether it is suitable to their subject matter, but simply because it is thought to be the only respectable way to do science. The philosopher John Dupré calls this "scientific imperialism," meaning "the tendency for a successful scientific idea to be applied far beyond its original home, and generally with decreasing success the more its application is expanded" (2001, p. 16). Once again, we see a frantic effort to generate models fitting the accepted paradigm, with little regard for the realism of the assumptions and mechanisms from which they are constructed.
In my mind this relates prettly closely to this post from Brad Setser (be sure to read the comments which wade deeper into modelling and risk measurement). The basic problems in risk management seems to me that assets are assumed to have inherent properties like the physical bodies. Particles have mass, position and velocity while assets have price, varience, and correlation. I call it a basic problem because these aren't really the properties of the asset but are the properies of transactions.

Transactions necessarily have assets, counterparties, prices and times associated with them but assets need not be part of a transaction. A series of transactions in two identical assets need not have a realtionship at all if the counterparties have no contact. The existence of exchanges and the requirments of open-outcry and disclosure get around this basic problem, creating attachment between transactions but I am not sure this does much more than create the illusion that assets always have prices and those prices move in continuous fashion.

No comments:

Post a Comment