Saturday, February 06, 2010

Garrison Orally On Reswitching

The following is from the Question and Answer session for Roger Garrison's recent lecture, "Macroeconomics: The Boom and Bust Cycle" (starting at about 59:29):
Audience member: "Can you reconcile using Hayekian triangles with reswitching and [unclear]?"

Garrison: "Yeah. The question is, 'How do you reconcile the Hayekian triangle with the Cambridge Capital Controversy, which involves capital-reswitching and technique reversing. I got that backwards - isn't it technique reswitching and capital reversing. Let me just take a poll here. How many have heard of the Cambridge Capital Controversy and know about technique reswitching and capital-reversing? Good, OK, good. One guy.

Let me just give you a standing-on-one-foot explanation. Capital theorists in Cambridge - this is Cambridge, England - argued that Böhm-Bawerk didn't have it quite right. His math was too simple, and if you allow for inputs and outputs to be expressed by polynomials, instead of rectangles or linear inputs, you find that the production process could lengthen or shorten with the change in the interest rate. If the interest rate falls, maybe it will shorten, or maybe it will lengthen. And you don't know. You have to look at the math and see how it turns out.

And the way the math turns out, there are only trivial exceptions - very small exceptions to the proposition that a low interest rate encourages long-term investment. You can come up with counterexamples, but even Cambridge capital-theorists, like Joan Robinson - she wrote an article, "The Unimportance of Reswitching" - just the magnitude doesn't amount to anything. If you read Hayek's Pure Theory of Capital, he acknowledges that, yes, yes, you can have funny profiles that seem to run counter to the general proposition that a low interest rate encourages long-term production. But those are very trivial. I've published articles on this. I can give you a link to one. But one thing I observed is that to get numbers where the reswitching is visible to the naked eye, you have to have the interest rate changing from 100% to 50%, or something like that, which itself suggests how minor that distinction is.

There are actually more fundamental criticisms that are based on how the Cambridge capital theorists measure roundaboutness. And the Austrians don't really measure it. They don't need to measure it. It's enough to show you the pictures. You saw the picture of the Research and Development chemist. You saw the picture of the retail clerk. That's what we are talking about, however you describe that mathematically. So I think, economically, that just doesn't work.

There are economists - Post Keynesians are the main ones - who throw up that Cambridge Capital Controversy as a killer of Austrian Business Cycle theory. They see that as the basis for just dismissing Austrian Business Cycle theory, but I think with very little justification."

  • Roger M. Garrison (1979) "Comment: Waiting in Vienna", in Time, Uncertainty and Disequilibrium: Exploration of Austrian Themes (ed. by M. J. Rizzo), D. C. Heath and Company
  • Roger M. Garrison (2006) "Reflections on Reswitching and Roundaboutness", in Money and the Markets: Essays in Honor of Leland Yeager (ed. by R. Kopl), New York:Routledge.


Ziragt said...

For all his published work on the topic, his response here seems to sum up his true feelings about the CCC problems for ABCT:
"It is empirically irrelevant; and, you know, aren't the triangles pretty convincing and intuitively appealing?"
Since both of these propositions can be pretty plausibly denied, where does that leave him?
Even more important, in my opinion, is a point Roger Koppl made about distinguishing between subjectivist elements in ABCT and physical capital and resources. It is unclear to me to what extent the theory relies on these two components and whether they are conflicting or not. Even if we can get around the CCC problems, this also needs to be addressed.

Anonymous said...


The only empirical study I have seen of reswitching estimated that it occurred 3-4% of the time. Do you have another study or estimate as to how often it occurs? I would consider 3-4% of the time as fairly irrelevant. (Their could be other reasons for why a small amount of reversal is relevant but I have not seen that argued.)


Interestingly enough I ran across a reference to capital reversal in Boldrin and Levine's totally neoclassical paper Perfectly Competitive Innovation.

On page 7 after the second indented equation:

the relation between Lt [labor] and either kt [capital] or xt[productive capacity] is ambiguous, as different
factor intensity rankings and the possibility of reversal may lead to a
non-monotone L(kt, xt). Rule out these altogether irrelevant cases by
assuming that T (kt, xt, L(kt, xt)) is increasing in kt, decreasing in xt,
and has a non-negative cross-partial second derivative.

I think Garrison's position is the general position of the profession as indicated by the Boldrin-Levine paper above. Reswitching and capital reversal is considered empirically irrelevant.

If heterodox economists want to make progress for the acceptance of their ideas, I think the critical point is showing the empirical relevance of reswitching.

Otherwise the neoclassical hammer may not be perfect but it is good enough.


BruceMcF said...

"I would consider 3-4% of the time as fairly irrelevant."

If this is the actual basis for the conclusion, clearly its a confirmation bias.

Bear in mind that the continuous curves of theory are not empirical curves, but are projections fitted to dispersed samples taken along the space.

A couple of reswitches located in the middle of the space that has been sampled implies that the fitted curves are guaranteed to be biased representations of the actual composites.

And assuming away the possibility in the econometrics means that all of the statistical tests of significance overstate the fit to the data.

In any scientific field, knowledge that a phenomenon that has been observed to exist "in the field" means that you are understating your likelihood of Type I errors would be something that has to be taken into account and corrected for, not something to be assumed away.

Robert Vienneau said...

I guess Kopl's point is that ABCT depends on commodities being assigned, objectively, to different orders. But if entrepreneurs have different subjective perceptions of technology, they will vary in the order that they assign commodities to. I agree, Ziragt, that Kopl has a point, although I am unsure how much importance I would assign to it.

I don't get Dan's point. I am aware of other empirical studies. To refute Garrison, I am inclined to cite Albin, Asheim, and Prince & Rosser. Furthermore, the Sraffian critique of the neoclassical vision of prices as scarcity indices does not depend on the existence of reswitching or of capital reversing. Anyways, I don't know why I should accept that empirical results are relevant to a point of logic.

Bruce, I had never thought before about how Han and Schefold's results were of the same order of magnitude as conventional levels of statistical significance.

Anonymous said...

You can come with counterexamples, but even Cambridge capital-theorists, like Joan Robinson - she wrote an article, "The Unimportance of Reswitching" - just the magnitude doesn't amount to anything.

It's been a while since I read that article, but wasn't Robinson arguing that the issues raised by the critique were far wider than just reswitching and that, as such, to focus on it was missing a lot of the issues? Hence its "unimportance"?

In other words, is quoting just the title somewhat misleading?

I'm going to the library this lunchtime (on an unrelated matter, namely anarchist community organising in Chile in the 1910s) but I'll have a look at the article and refresh my memory...

An Anarchist FAQ

Anonymous said...

I re-read (quickly) the Robinson article and her point was that the CCC had been sidelined into discussing reswitching as if that were the be-all and end-all of the critique. Her basic point was that the neo-classical economists had not answered her point yet.

I'm not sure how invoking that article addresses the issue that Robinson (and others) was raising. Maybe I missed something, but the article was about trying to refocus the debate onto the real critique (in her eyes).

"And the Austrians don't really measure it. They don't need to measure it."

So the whole thing is based on something you cannot measure... and so, then, completely up in the air. If you cannot measure it, you cannot refute it empirically and so the ABC is built in such a way that no matter the criticism it can be ignored... Ah, the joys of science!

"They see that [the CCC] as the basis for just dismissing Austrian Business Cycle theory, but I think with very little justification"

Oh, there are plenty of reasons for dismissing the ABC -- its use of "natural" rates of interest (i.e., equilibrium), its assumption that it is the nasty government which somehow forces banks to expand credit rather than profit seeking, Kaldor's critique from the 1930s, Sraffa's critique, and so on.

All in all, the ABC seems fatally inconsistent with the other axioms of "Austrian" economics which pays lip-service to dis-equilibrium, glorification of entrepreneurial profit seeking, opposition to state regulation (how else are they going to enforce a 100% gold reverse policy? The market has never done so...) and such like.

The ABC appeals to those who seek to blame everything other than capitalism for its business cycle. It is an ideological solution to a political problem (the difficulty of glorifying a system which regularly goes into crisis). As would be expected, it starts to come apart when looked at closely.

An Anarchist FAQ

Anonymous said...


First, thanks for the link to the other studies I will look at them.

(Note, in this post, I am only concerned with the reswitching issue and not the broader relevance of sraffian economics.)

Anyways, I don't know why I should accept that empirical results are relevant to a point of logic.

My point in citing Levine and Boldrin is that yes the logic is accepted. Everyone accepts the logic. But a logical possibility does not make it empirically relevant. That is also my take away from what Garrison is saying.

Sure reswitching is possible, just like its possible that I might win the lottery. But I am not going to live my life with the expectation of winning the lottery. And at the moment, I don't think economists would benefit from worrying about issues like reswitching.


Robert Vienneau said...

Iain, I agree with your point about Robinson's article. It is not about the quantitative magnitudes of anything associated with reswitching, and Garrison's cite of it is unjustified. Robinson's article is about the need to analyze processes in historical time. A comparison of stationary states is not such an analysis.

Other of Garrison's statements are equally mistaken. Cambridge theorists in either Cambridge don't need to measure roundaboutness at all. In a capital-reversing example the cost-minimizing technique at a higher interest rate can have greater output per person-year in examples where both net output and labor have natural physical measures. All of Yeager's hand-waving about "waiting" cannot make such examples consistent with the Austrian vision. Furthermore, I don't see how Hayek's abandonment of his triangles justifies Garrison's retention of them.

Anonymous said...

How can an Austrian economist serously claim that reswitching is empircally unimportant?


Levi said...