Wednesday, August 14, 2013

Pete Boettke asks why ABCT isn't on top

Here. The answer to that is pretty complex. If I were to provide a full answer I'd probably reject a lot of his premises in the post too. I don't have time for a full answer write now but I did write up part of my answer on facebook, which I'm reproducing in full below:
I think a lot of it is path dependence in the science and it need not stay that way. Indulge me in a stylized history of thought as to what that path dependence consisted of... I promise I'll get back to ABCT by the end.

Keynes did a better job of explaining the depression so he came out of the 30s and 40s on top. There were informal microfoundations in Keynes and the same worries about microfoundations that Lucas would later express. This is true of Hayek too - there are informal microfoundations but nothing particularly formal in the presentation of ABCT (I've only read portions of Pure Theory of Capital... maybe there's a lot more in there but since Austrians even seem to get headaches from that book perhaps it hasn't been the best ambassador for ABCT).

When Lucas (and Phelps) pushed the microfoundations point Keynes was dominant so of course what got reformed was a microfounded Keynesianism which is essentially the consensus macro model today.

ABCT was not dominant when the Lucas critique came around so ABCT never got the formal microfoundations that modern economists expect of a theory. In that sense I'd kind of take the opposite view of yours in the post - even though there's informal discussion of microfoundations, it's the lack of microfoundations in ABCT that is it's major liability right now.

This need not be the case. ABCT explores a lot of interesting issues around the capital structure that no one else is really talking about. I'll have to take a look at the Calvo article and how it treats the problem.

Recently I wrote a post on what I think is the best way forward for modeling ABCT in a way that mainstream economists would find plausible.

In a nutshell, I think the Romer endogenous growth model is structurally the most similar to Hayek's. Currently in the Romer model there is no time component to intermediate goods. Adding a time component to tie in the interest rate to decisions in the intermediate good markets and then doing an analysis of the impact of interest rate changes should reproduce the essential elements in ABCT in a very highly regarded mainstream macro model.

I'd try it myself but it's tangential to my current dissertation plans.... maybe IHS can incentivize me with some money :)

One other probably much more obvious contributor is that Hayek dropped a lot of his macro work after the 40s. That SURELY made a dent in the success of ABCT. Imagine if Hansen and Samuelson and Tobin and Patinkin and Robinson all just said "oh well - I'll talk about macro from time to time but I'm going to do my really innovative work on other problems"... Keynesianism wouldn't be in very good shape.

Treating Keynesians like they don't understand basic economics probably hasn't helped ABCT's prospects either."


  1. "Keynes did a better job of explaining the depression so he came out of the 30s and 40s on top. There were informal microfoundations in Keynes and the same worries about microfoundations that Lucas would later express."

    With all due respect Daniel Kuehn...haven't we gone over this before?

    There *IS* a supply side to The General Theory.

    There's a reason why J.M. Keynes targets A.C. Pigou's 1933 book, The Theory of Unemployment. J.M. Keynes even says in the Appendix to Chapter 19 of his magnum opus that there is "no significant difference" between his and Pigou's "modes of expression". If you want to go to the specific page in The General Theory, I'm quoting from Page 273.

    Furthermore, J.M. Keynes does make it fairly clear (albeit only briefly, like on Page 24 in Chapter 3, though he does discuss it also in Book V) that he keeps the behaviour of firms in his model the same as one would find in Principles of Economics by Alfred Marshall: the profit-maximizing entrepreneur or businessperson. You can also gather from the book that J.M. Keynes uses perfect competition/pure competition as one of the building blocks in The General Theory.

    I apologize for coming off as flustered, but I think you need to get a copy of The Theory of Unemployment by A.C. Pigou and reread Book V of The General Theory. Also, please take the time to digest the following sources.

    Brady, Michael Emmett. "Keynes, Pigou and the supply side of the General Theory." History of Economics Review 21 (1994): 34-46.

    Brady, Michael Emmett. "A Study of J.M. Keynes’ Marshallian–Pigouvian Elasticity Approach in Chapter 20 and 21 of the GT." History of Economics Review 24 (1995): 55-71.

    Brady, Michael Emmett. "A Comparison-Contrast of J.M. Keynes’ Mathematical Modeling Approach in the General Theory with some of his General Theory Interpreters, especially J.E. Meade." History of Economics Review 25 (1996): 129-159.

    1. I don't get why you are so "flustered". I don't even get the point of this post. Did I ever say there was no supply side? Did I ever say there were no profit-maximizing entrepreneurs?

      The point is, are these formally or informally discussed? Informally for the most part. It is not microfounded in the way that modern models are microfounded.

    2. Well, perhaps this is a tempest in a teapot, and I was wrong to make my comment. Would it be safe to say that J.M. Keynes's microeconomic foundations in The General Theory are not as up-to-date as you would find in a journal article by say, Roger Farmer or Christina Romer? Perhaps.

      I would agree with you, Daniel Kuehn, that J.M. Keynes does not model "animal spirits" in a formal way - this is where Daniel Ellsberg's 1962 doctoral thesis (Risk, Ambiguity and Decision) improves upon J.M. Keynes. In Ellsberg's doctoral thesis, there is an "optimism-pessimism index" that can capture such crowd psychology in the market.

      J.M. Keynes's choice of equations is differential calculus and integral calculus. Perhaps this isn't as "rigorous" as using the Calculus of Variations, but it is still something that can be considered formal modeling. But it was state-of-the-art for the time.

  2. "[Keynes] keeps the behaviour of firms in his model the same as one would find in Principles of Economics by Alfred Marshall: the profit-maximizing entrepreneur or businessperson."

    But that's not microfoundations in the sense that somebody brought up on Hal Varian's textbooks would expect. Keynes's entrepreneur takes (W,P) as given, despite the fact that there are unemployed workers out there; why doesn't he cut a deal with them? Keynes has an explanation of sorts but it isn't microfounded.

    1. Perhaps it isn't "microfoundations in the sense...[of] Hal Varian's textbooks". But to say that there are no microeconomic foundations or no formal model in The General Theory is incorrect. Please see the aforementioned articles published in the History of Economics Review, Kevin Donoghue.

    2. A little math does not a formal model make, I don't think. There was math in the General Theory that points you in the direction of the model that Keynes has in mind. There were not formal microfoundations as the word has been understood since it was first used. There's not even a compact macro model that you can point to as his "model" - you're piecing together lots of different things even in that case.

    3. I'll be succinct here for now...then why are there multiple references to Book V of The General Theory in footnotes throughout his magnum opus?

    4. Because it's an important book where a lot of the formal statement of ideas are developed and because it puts his ideas in the context of Pigou's book. So?

      You seem to think people are arguing that there is no theory whatsoever.

      There is. There is just not a modern microfounded theory in the modern sense. There are lots of informal discussions of what we would call microfoundations.

    5. Well, I'm glad you get the point of the comparison-and-contrast with A.C. Pigou's 1933 book.

      I'm aware that you're not saying that there is no theory whatsoever. What I do disagree with however, is the description of Keynes's microfoundations as being "informally" discussed.

      Keynes uses the standard marginal productivity theory of wages (with one variable input - labour - and one fixed input: capital) in his magnum opus (despite making a brief criticism of it in a footnote on Page 140, that's what he - and A.C. Pigou - both use). That can be considered a formal component of his microfoundations, even if it isn't as complex as the formulations by say, Hal Varian.

      I would agree with you that it isn't as rigorously expressed in formal terms as the microfoundations of modern times are, but I don't think that you can call Keynes's microfoundations "informal". Please see the following piece for a longer discussion.

  3. Daniel, I've spent a bit of time thinking about your idea how to model ABCT (ah, the joys of procrastination...). It seems to me that modifying Romer model would lead to some analytic results about the steady state, but incorporating uncertainty and dynamics may be tough. Perhaps some kind of discrete-time RBC model with multiple sectors and time to build, with different production time across sectors, could be a better approach (of course then one would still have to model monetary policy somehow).

    1. I've never worked with the dynamics in the Romer model - not sure how tough that is. Uncertainty, I would think, is a minor issue in formalizing ABCT (or I should say bringing it up to modern standards of formalization). Certainly comparative statics will make the point nicely regardless of what one chooses to do with the dynamics.

      Your idea sounds good too. I only have minimal experience with both RBC, and time to build was only a brief side discussion, where I've worked a lot more with endogenous growth models. But that's just my comfort level in thinking through something. Time to build obviously incorporates a lot of these issues. The only hesitation I'd have is that the sectors would have to be associated with the time structure, right? You don't just want a bunch of horizontal sectors with potentially different times to build. Instead you want a bunch of vertical sectors connected through time. I don't know how adaptable time to build is to that, but the Romer model seemed natural to accomplish it.

    2. What I meant is that if the economy uses capital goods with length of production say 0 to 2 years, then state of the economy at time t will have to include distribution of current capital goods "under production", which in turn will depend on whole history of aggregate variables (wages, interest rates) from t-2 to t. Modelling this in continuous time sounds hard.

      I'm not quite sure I understand what you mean by difference between vertical and horizontal sectors. I was thinking about something like this: final good is produces from several capital goods. Capital goods are produced using labor, and capital good of type i takes i periods to complete. If the production function for final good is symmetric, output is maximized when equal amounts of capital goods are used - but if producers discount future profits, sectors with longer periods of production will face de facto higher marginal costs, and thus less such capital goods will be employed (compared to equilibrium without discounting). Thus if households were more patient, interest rate and the "distortion" from symmetric outcome would be lower and output higher (at least in steady state).

      Anyway, my knowledge of ABCT comes mostly from reading bad arguments on libertarian blogs, so I'll let others try :)


All anonymous comments will be deleted. Consistent pseudonyms are fine.