Reply to a Economic Theorist of Integrity, on the Scandal of Graduate Programs in Economics

Thanks for your letter. My impression that you are a scholar of integrity is thus confirmed!

I note that you, like most mathematical economists who have thought about it a little, agree with me that statistical significance is bankrupt. I wish you would challenge your econometrician colleagues on the matter beyond an occasional sharp remark about How Big is Big in seminars. There's an implicit non-criticism agreement between theorists and econometricians (in place since Koopmans first articulated it long, long ago): you let me play pointlessly with existence theorems and I'll let you play pointlessly with significance tests! We should break it. The first-year programs in econometrics, as you might well agree (Jim Heckman and Orly Ashenfelter say they do), have been ruined by mechanical reliance on t-tests and unit-root tests. Our students never learn how to do serious empirical work, only how to read SPSS output. As I wrote to you in my first letter, I was charmed that Ken Arrow has agreed with me on the point for many years (he said it first in 1957, for example. . . when I was 15 years old!).

But Ken did not agree when I claimed at a big session of the AEA we had some years ago in Chicago that economics departments have massively overinvested in what is called Theory (that is, mathematical economics taught in the style of the Department of Mathematics instead of the Department of Physics), and that the first-year programs have been ruined by Mas-Colell-type teaching of micro, because the Theorists have been assigned to teach it. You, for example!

The gap between my view and yours (or for that matter Ken's) is a difference between us is what is meant by "theory," and in particular what is meant by that little yet powerful word you use, "tools." We entirely agree that students need to learn to think like economists, pronto. No mercy. Tough stuff. Get used to it. This term. Learn the tools of an economic scientist. You need them for all your work as an economist.

What has gone increasingly wrong, though, with the teaching of graduate theory since Samuelsonianism took hold, about 1960, is that the only "tool" the students have been taught under the heading of "theory" is Max U under constraints. They do not get taught how to simulate with variant functional forms or with non-analytic solutions (which is the main tool of physicists and engineers, for example). They have not been taught how to argue verbally to an economic conclusion (which is the main tool of historians, though often mishandled when they are talking about the economy). They have not even been taught the main tool of economics in the mid-20th century, geometrical argument. They do not know the history of economic thought, and therefore for example they reinvent monopolistic competition or the theory of second best every ten years or so (cf. Paul Krugman). They have no idea, indeed, that there are forms of mathematical thinking (network theory, nonlinear dynamics, topology other than fixed points) having nothing to do with constrained maximization that might nonetheless be scientifically fruitful ways of thinking about economic issues. They certainly do not learn that storytelling and metaphor-making are means of thinking that may or may not have mathematical expression yet are essential to the formation of an economist.

So a Mas-Colell-type course teaches the first-year student how to use one tool, a hammer, and then declares that the whole world is a nail. You complained in your reply to me about how badly macro-economists did in predicting the Great Recession. (I myself think that prediction of this character, which would imply unlimited opportunities for personal profit, is anyway impossible, and that what was wrong in the macro-economists is that they didn't know any economic history.) You should consider what Krugman said in August in the Times Magazine (though he showed there by the way that he, too, doesn't know a thing about history): that the models were wrong precisely because they were Theory of the sort you admire.

Furthermore, the usual first-term graduate course in Theory teaches from day one that proofs of existence are relevant to a science. They are not, as one may learn by asking your local physicist (or biologist or historian, for that matter). I explain in various writings why existence theorems are philosophy, not science (if you want the evidence directly from physics, start reading Richard Feynman's popular and teaching works):

• Knowledge and Persuasion in Economics (Cambridge 1994)
• The Vices of Economists, The Virtues of the Bourgeoisie (Michigan/Amsterdam 1997)
• The Secret Sins of Economics (Chicago 2002), available free on the internet

and in various articles, available now on my web site deirdremccloskey.org, for example:

"Economic Science: A Search Through the Hyperspace of Assumptions?" Methodus 3 (1, June 1991): 6-16. Reprinted as pp. 73-84 in Craig Freedman and Rick Szostak, eds., Tales of Narcissus--The Looking Glass of Economic Science, New York: Nova Science, 2003.

"The Trouble with Mathematics and Statistics in Economics," History of Economic Ideas 13 (3,2005): 85-102, delivered to MUIR-PRIN project The role of mathematics in the history of economics, Venice, January 28, 2005, with replies by Dardi, Egidi, Marchionatti, and Fontana.

My experience with other economists who like you say that they have "strong opinions" about methodology is that they have not read such works (or indeed much of anything in epistemology aside from the amateurish reflections of economists on the subject, such as those by my former colleague the late Milton Friedman or by my mother's mixed-doubles tennis partner the late Paul Samuelson). In particular most economists have not faced up to what I call the "A-Prime, C-Prime Theorem": namely, that any conclusion can be "justified" with a Max U theory (or any style of theory, for that matter) if one is free to select the Assumptions, A. Slight perturbations in assumptions reverse the conclusions, C. It is an analytic truth. But it does simulate pretty well the history of most theoretical controversies in economics, from Ricardo's views on capital deepening to the latest wrinkle in solution concepts or trembling hands in game theory. One set of assumptions A map into a set of conclusions C, such as convexity implying that free trade is good. Modify convexity, A-prime. Therefore (the "policy conclusion" tacked on to the end of a paper in trade theory) free trade is bad. Introduce a new assumption, A-double-prime, about internalization of externalities. Free trade good. Introduce a new assumption, A-treble-prime, about monopoly. Free trade is bad. And so forth, to no scientific conclusion.

It's philosophy or math-department math or theology — wonderful fields that I much admire (really: I've written numerous books and articles about and even in them), but none of them is the slightest bit scientific. They are about thinking, not about the world. Science is about the world. The unbounded hyperspace of assumptions has been explored under questionably relevant factual side conditions. So scientifically what? I'm not "against theory," meaning disciplined thinking about the economic world. I'm against Theory viewed as a pointless game of assumption-varying confined to Max U with no quantitative discipline relevant to the economic world (forget about econometrics in its present state as constituting such a discipline!).

Having taught maximization under constraints myself for many years (my old The Applied Theory of Price is available, too, on the website), I of course agree that Max U has his place. He's a sociopath, admittedly, but in much of our anonymous behavior all of us are, sometimes, and lamentably, focused on indulging our own "tastes" without other ethical considerations. Garrison Keillor the comic writer and entertainer notes in the newspaper today (Chicago Tribune, January 14, 2010) that when driving among strangers even he, a nice Lutheran boy from Minnesota, swears sociopathically at people who cut him off. One is going to learn more about covered interest arbitrage in the markets for dollars and euros and yen by assuming Max U behavior than by appealing to the sense of justice or love of the dealers.

But we should be teaching young economists to be economic scientists, not merely constrained manipulators of Max U. Among the things we should be teaching them is to be ethical as scientists (often enough they draw the conclusion from Max U that a scientist should be a sociopath, too). That entails searching for insight into the economic world wherever it may be found, in novels as much as in Lagrange multipliers, in their own daily experience of an economy as much as in the pages of the Journal of Economic theory. Yet I see that in your own teaching that you are starting the students off in an ethical way, for which I commend you. You are giving them at least a brief example, even from that witch McCloskey, of improving criticism of the sad folk ways of our science.