1. Skip to navigation
  2. Skip to content
  3. Skip to sidebar
Source link: http://archive.mises.org/1768/realism-and-abstraction-in-economics/

Realism and Abstraction in Economics

March 29, 2004 by

Realism and Abstraction in Economics: Aristotle and Mises versus Friedman by Roderick T. Long (Auburn University)

Austrians have frequently criticised neoclassical economics for the unrealistic character of its assumptions. Neoclassical models are typically “idealised”; that is, they leave out such features of the real-world economy as rivalry, imperfect information, non-monetary incentives, and the passage of time. In his enormously influential 1953 article “The Methodology of Positive Economics” – a work which Friedrich Hayek once described as being “as dangerous” as Keynes’ General Theory – Milton Friedman defended the use of unrealistic models against Austrian-style criticisms, on the grounds that any good explanatory theory must be abstract, and abstractions by their very nature are unrealistic. [MORE]


amcguinn March 29, 2004 at 9:17 am

I think the distinction here is _entirely_ between empirical and deductive reasoning.

The Friedman position is the scientific one. He does not, as Long says, assume that knowledge is perfect when it isn’t, he assumes that the results are the same as they would be if knowledge were perfect.

In the scientific tradition, what that assumption has produced is a _hypothesis_. It is a useful one if it is confirmed, but there is no reason to believe it true until it is confirmed (repeatedly), and it is subject to being falsified by experience.

What Long is arguing for is a deductive approach, philosophical or mathematical rather than scientific, where a theory rests on the reasoning that produced it, rather than observed confirmations of it. Indeed, “predictions” such as “unemployment will be higher than it otherwise would have beeen” are impossible to confirm by observation.

Which is the right approach for economists? Neither has been very successful thus far. The word that is missing from Long’s description of the deductive approach is “axiom”. The answer to the question on page 20 regarding disagreeing mathematicians is that they check each others’ reasoning from axioms to conclusion. If they don’t agree on axioms they have no basis for reaching agreement. Mathematicians in Arisotle’s day, and very nearly in the present, agree on axioms. I’m not sure that the same can be said of economists.

amcguinn March 29, 2004 at 9:50 am

Further to the above, I see that the author does deal with the subject of axioms, via “presuppositions”, in the referenced essay on Collingwood at http://mises.org/asc/2003/asc9long.pdf . I don’t feel that the “Realism and Abstraction” piece is able to stand apart from that discussion.

Roderick T. Long March 29, 2004 at 1:02 pm

I don’t agree that the distinction I talk about in my paper “is _entirely_ between empirical and deductive reasoning.” It’s related, because one needs the precisive/non-precisive distinction to explain why the a priori method can be trusted in economics. But what I was pointing out is that the laws of *physics* are also best interpreted as non-precisive rather than precisive abstractions — and the laws of physics are *empirically* grounded. (That’s why I didn’t talk about axioms in that paper.)

Hence I also don’t agree that counterfactual statements are “impossible to confirm by observation.” If that were true, it would throw all of physics out the window, since all laws of nature either are or entail counterfactuals (that’s the difference between causality and coincidence).

As for whether I misdescribe Friedman’s position — I simply describe it the way Friedman does. Of course Friedman doesn’t assume (in the sense of “believe”) that knowledge is perfect, but he does advocate using models that assume it, and he thinks — mistakenly, I argue — that in doing so he is imitating physics.

Incidentally, my chief defense of apriorism is not the Collingwood piece but the Wittgenstein piece.

Comments on this entry are closed.

Previous post:

Next post: