Technique Is Not Wisdom

by Don Boudreaux on October 14, 2009

in Economics

Here’s a letter that I sent yesterday to the Wall Street Journal:

David Henderson rightly laments that “that mainstream economics has become highly mathematical and increasingly independent from reality” (“A Nobel for Practical Economics,” Oct. 13).  And he also rightly hopes that the award of this year’s Nobel Prize in Economics to Elinor Ostrom and Oliver Williamson – economists who “do the time-consuming work of examining the institutional structures that humans build to solve their own real-world problems” – will help to break the profession’s enchantment with formal mathematical modeling.

That this enchantment is overly strong is revealed by William Baumol’s 1986 review of Williamson’s magnum opus, The Economic Institutions of Capitalism.  Although Baumol praised the book as a path-breaking contribution, he regretted that it offers no formal models.  But, observed Baumol, the “absence [of formal models] is not to be ascribed to the inability on the part of the author, who in the past has shown himself to be a master of the pertinent techniques.”*

There you have it.  Williamson’s book is wonderful – and apparently more trustworthy because its author knows math.  But, by golly!, wouldn’t the book have been even better if it were formalized?

I, for one, learned far more from any one of Williamson’s non-formalized works than I have from any 100 formal models.

Sincerely,
Donald J. Boudreaux

* William J. Baumol, “Review: Williamson’s The Economic Institutions of Capitalism,” Rand Journal of Economics, Vol. 17 (Summer 1986), pp. 279-286.  The quotation in the letter is found on page 285.

Comments    Share Share    Print Print    Email Email

  • pauljohnson
    Another interpretation is that Baumol is writing for the readers of the Rand Journal and knowing his audience feels he has to address the math issue: don't be put off by the absence of formal models - this is a great book!
  • Manfred
    One has to be careful in critizing models. Models can help quantify effects, can help sort out logical fallacies and logical mistakes, and can help with the comparative statics. Keynes' famous book [his "General Theory"] was written without a single equation, and the profession spent the next 50 years sorting out "what Keynes really, really, REALLY meant".

    Yes, formalism can be abused, and one really wonders if every paper published in the Journal of Economic Theory or even Econometrica has value from an econ point of view. Even the American Economic Association, a few years ago, had studies done by very prestigious members of the profession [Debreu, etc], that discovered that grad students in the best graduate programs [Harvard, MIT, Chicago] could solve a highly complicated general equilibrium problem, but could *not* answer relatively "simple" questions like what happens to the price of a hair cut when the price of scissor increases.

    Again, formalism can be abused, yes, but dismissing modeling outright, I think is a mistake.
  • danielkuehn
    RE: "Keynes' famous book [his "General Theory"] was written without a single equation, and the profession spent the next 50 years sorting out "what Keynes really, really, REALLY meant"."

    Sure it did! Chapters 6, 8, 10, 14, 19, 20, and 21 have equations in them. Chapter 23 has poetry and Chapter 24 has philosophy. What's not to like? :) It's true he doesn't rely on a lot of math, but it's definitely in there.

  • Manfred
    Stand corrected. Thanks. M
  • danielkuehn
    RE: "Williamson’s book is wonderful – and apparently more trustworthy because its author knows math."

    He never said or implied this Don.

    Good economics rests on three pillars: informal intuition, formal modeling, and empirical verification. Any insight is stronger when it has all three. You act like people who appreciate these pillars just want math for math's sake.



  • william_sjostrom
    I think you are missing Baumol's point. Just before the line you quote, he writes: "If there is any one approach in whose use members of our discipline excel relative to those in other fields, it is in the construction of formal models that readily yield illuminating theorems and that lend themselves to econometric testing." Baumol's point, which he elaborates for nearly two pages, is that formalizing the model would produce better comparative statics, and better econometric testing. Not, as you put it, math for the sake of math. It is no doubt true that formal modeling has had excessive prestige in the profession for a long time, and the '80s and '90s were particularly bad for it. (I suspect that Baumol's remark about Williamson's formal abilities was related to Rand being at the time a hotbed of useless articles doing yet another useless twist on game theoretic equilibria.) There are a depressing number of articles going from "I have a neat theoretical result" to "public policy should therefore be changed" without any evidence in between. With a few exceptions, such as the Penn World Tables, gathering data carefully is a hugely unrewarded activity. Steve Cheung loved to tell about a fisheries conference he was invited to after he published his 1970 paper on non-exclusive resources. It was in a hotel overlooking the Pacific, and at one point everyone got really excited because a fishing boat sailed by, and most of the theorists had never seen one before. So yes, good economics is a lot more than formalized theory, which Baumol explicitly conceeds. But Baumol argues (I think correctly) that formalized theory is important for advancing work from ideas to full scale testing.
  • This quote from Baumol is hilarious.
  • Don,
    I think that the the issue of the excessive formalization of Economics is a red herring. Tell me how to measure the contribution of a paper to the discipline, and the right amount of Mathematics will sort itself out. By focusing on technique, you are criticizing the use of a rather agnostic input, rather than the much more relevant output.

    Successful disciplines do not even have this sort of conversation: in mainstream Statistics for example (as in Statistical Science or Journal of Am.Stat.Soc.), Mathematics is often used in a relatively discursive manner. An approach can be presented without a formal proof and still be of great interest. In Physics, you can run the gamut of techniques, but a lot of important theoretical Physics is not mathematically rigorous. These disciplines have clear criteria of value however, and these keep the excessive theoreticians in check. Their work is sterile, and it shows quite quickly.

    The editors-in-chief of the major economic journals should ask themselves what their subject matter is, and how do they measure contributions to it.
  • This flirts with false dichotomy territory. Williamson's book may well have been informative without formal models, but is there any reason to think it would have been any less informative with them? More likely if it were the same book, plus formal models, it would be a clearer work than it already is. If that is Baumol's point, then I agree with him in general. Formal models are no excuse for slacking on verbal clarity, but in general I find academic work easier to understand when it is formalized.
  • danielkuehn
    I agree - I think this is Baumol's only point, Don. And people since Williamson have formalized the New Institutional Economics.

    Valuable economic insights rest on three pillars: informal intuition, formal modeling, and empirical verification. They aren't opposed to each other, as a lot of people have been suggesting lately. They support each other. And we should prefer a theory or a body of literature that has all three. If Williamson presented the intuition and formal model to provide framework for that intuition it would have been "better economics". Maybe not as accessible or as quick to take hold, and that's important to consider too. I think it's dangerous to make intuition, mathematical models, and empiricism into enemies of each other.
blog comments powered by Disqus

Previous post:

Next post: