Friday, October 3, 2008

What Would a Scientific Economics Look Like?

This is the title of piece of mine just published in the Post-Autistic Economics Review. I happen to be a fan of science (I’m so pre-postmodern) and would like economics to move in that direction. Let me know what you think.


Anonymous said...

i was a big fan of science, but due to "the big crunch" in which the hubble (or bubble) expansion halts, i have gone back to 'natural philosophy'; is it scientific to turn a trick, or a postmodern treat?
(see T. Andresen on 'economic black holes" among others).

i have some problems with 'post autisitic econ' (similar to Gintis), but its probably no worse than classical (or neoclassical for those who believe you need time for money, becuase of Keynesian friction----a theoretical absurdity because in a quantum world one has frictionless motion in the imaginary ('utopian socialism' or 'mutual AIDS').

probably the people who did the classical orthodox stuff were 'garbage' (ARROW, SAMUELSON, ETC.) but if one simply reads them without social context there is no problem. Or, they are no worse than Boltzman''s ideal gas as a 'start' or first approximation. (The SMD theorem to me seems actually quite consistant with GET, though maybe not the way most people portray GET. Newtonian 3 body problem is similar (see D. SArri). This is similar to the way A Smityh is presented by some libertartian types (the great Stepehen Moore of the Club for Growth). Behavioral finance really is just a method to estimate the utility function rather than being some 'new contrarian thing'. (The problem with it is like with GET or PAE---the idea you can use 10 grad students from your course, or 10 special 10 years olds, to estimate this is BS (but one could say, like the ideal gas, a start. ) CGE modeles seem common sense (just i/o tables eseentially) like a budget; sure they are identities but likely some accounting is useful and neccesary (e.g. 'wrath of god'.)

The PAE types throw alot of words where equations (were once) used to lie, but its unclear whether this isn't just like a Crips alternative to the Bloods.

So its possibly the issue is partly who does the economics, as opposed to the content.

Maybe people could try to actually read the journals to see what is scientific, and what is lacking. Especially rather than creaTING NEW ONES.

Type 1 errors, mechanisms, lip services, tricks.

Economics is so value laden its almost impossible to make it scientific. (If we follow Chomsky and other rational martians then hopefully we can seed other planets using the Siren of Titan to actually do some good experiments). Are there absolute values or are there heterodox preferences generated from one superstring (or E8)?

Besides fish, there is 'evolution' (a theory---its on wikipedia). While theoretical biology (eg Rashevsky, Lotka...) can be some frutless search for a theory of maximum power in principia, field biology can be just a collection of anecdotes. There is probably too much theory and data, much of it useless. The ants are still searching for sugar.

i noticed that too about the 'identities'. a good point. alot of those identities are used to pretend their is some mathematical derivations going on (i think the parecon book is like this but its probably true for AER and JPE where they include a few hyperbolic discounting terms in the bellman optimal control problem for aesthetic reasons.)

but then math is essentially a tautology or identity, and even evolution as compared to fish is not a process but the tautology that 'survivors survive'. said...

[Just back from Europe]


You don't say too much about it, but of course the experimental economists claim that this is what they are all about. However, there is a major explosion of arguments erupting among them about their various methodologies, much of which will be appearing in print over the next year or two.

Peter Dorman said...

Welcome back, Bark. I do mention experiments in passing at the end, and I think this is a potentially important development. Nevertheless, there are two big issues that experimental economists have to grapple with, in addition to all the other, techier stuff. One of course is external validity; we all know that. The other, which is implicit in my article, is whether the experiment is designed in such a way as to capture mechanisms, or whether it is solely about end results.

Anonymous said...

i glanced at an older article on the same theme by dorman in the review. that seemed more understandable---the distinction between mechanisms and end results i associate with 'path dependence'. however, i think behavioral finance and game theory both may be falling into the same trap as general equilibrium theory---the first is often based on somewhat rigged and possibly trivial formats so that it is hard to know what general signifiance results have. (this reminds me of the current situation with finding genetic markers associated with behaviors; often they use small samples and are not reproducible; same with iq and education research). game theory is highly formal and ends up studying toy models which are often ad hoc variants on a theme---almost like anecdotes. (Sunnstein's book 'nudge' which i haven't read i think does capture the basic gist of the behavioral approach, though so did bounded rationality. Sunnstein often seems wrong however on policy, even is he has a good understanding, like sachs and krugman.)

speaking of krugman, his article on ricardo on his web site also discusses path dependence and multiple equilbria; however at the same time he dismsisses the santa fe school (eg arthur) as being fairly irrelevant; i view them as possibly more elegant reformulations of the same ideas (also possibly played out or in need of a transfusion). krugman even had a paper i think discussing symmetry breaking, which didn't reference physics.

i wonder what people thought of ormerod's complaint in the same journal about peer review; his models often seemed fine, but unless the data anlyses is new ideas about contagion are fairly standard.