May 2nd, 2010

Word of the Day: Ergodicity

I'm reading _Econned_, reviews of which from numerous sources lead me to at least provisionally trust it. At or around location 945 (10% of the way through the text on the kindle), I encounter this assertion:

"One way to assess mainstream economics is to look at the assumptions and ascertain whether they make sense. One of Samuelson's major assumptions that undergirds much of mainstream economics, and one that Keynes rejected, is ergodicity.

In simple terms, "ergodicity" means that no matter what happens in the world, everything will reach a point where things stop cange, which, in economics, is the prized equilibrium. It also means that a system acts consistently over time."

This is a quite startling assertion, however, googling around indicates that it is actually a valid summary of the idea of ergodicity in economics. I find this utterly amazing, given the source of the idea in physics, where the people who came up with the idea (not in economics!) certainly understood that it is just not safe to make this assumption without a lot of very careful observation, and making sure that observation was detailed enough over a long enough period of time.

Not Just a River in Egypt, also, the Hippo in Flight

Again, from _Econned_, this time 19%/location 1785ish, Paul Cootner is quoted on the subject of Mandelbrot's work:

"If he is right, almost all of our statistical tools are obsolete. ... [that's in the book, not from me] Almost without exception, past econometric work is meaningless. Surely, before consigning centuries of work to the ash pile we should like to have some assurance that all our work is not truly useless. If we have permitted ourselves to be fooled this long into thinking the Gaussian assumption is valid, is it not possible that the revolution is similarly illusory?"

There's more in _Econned_ and presumably more at _The Random Character of Stock Market Prices_, published by MIT Press in 1964. Page 337.

(Judging by what I turn up via google books, what was omitted in _Econned_ is: -- "least squares, spectral analysis, workable maximum-likelihood solutions, all our established sample theory, closed distribution functions.")


It is to laugh.

The hippo is the hypocrisy of a member of a discipline which makes it a point of dogma to discount sunk costs to zero engaging in this particular flight of rhetoric.

Of course Mandelbrot was right; besides, only a moron would have ever accepted the Gaussian assumption in the first place. Did these people pay _any_ attention to actual markets? Like, ever?