Thursday, August 10, 2006

Stiglitz: Competitive Economies Almost Never Efficient

From Steve Keen's debunking economics page, I find that Stanley Alcorn and Ben Solarz report on "The Autistic Economist" in the Yale Economic Review. A couple of quotes:
"The First Fundamental Theorem of Welfare Economics [asserts] that every competitive equilibrium is efficient. Based on the work in information economics for which he won the Nobel Prize, Stiglitz finds that this most basic claim is not robust to the removal of the assumption that information is perfect... 'Quite contrary to that theorem', Stiglitz concluded, 'competitive economies are almost never efficient.'"

They also quote Stiglitz as saying that "[Economics as taught] in America's graduate schools... bears testimony to a triumph of idology over science." According to the authors:
"Economics courses at the undergraduate level typically place little-to-no emphasis on learning the tools of economic science, instead focusing on teaching algebraic simplifications of actual economic work, and then assigning problem sets in which students plug in values for the different variables. It is good practice, perhaps, for a few specific mathematical techniques, namely constrained maximization, but it is hardly a training in how to think creatively about dealing with the economy. The bedrock of economics as it is taught is not the subject matter - the economy - or even the approach - the neoclassical school of thought - but ideology, as Stiglitz said. The repetition of simplified and vulgarized economic conclusions is the main task of introductory, intermediate, and even some advanced economics courses, and little else sticks with the students."

I haven't read much of Stiglitz on information asymmetries. I do have this article handy:
  • David M. G. Newbery and Joseph E. Stigliz (1982). "The Choice of Techniques and the Optimality of Market Equilibrium with Rational Expectations", Journal of Political Economy, V. 90, N. 2 (Apr.): 223-246

2 comments:

radek said...

”Stiglitz finds that [The First Fundamental Theorem] is not robust to the removal of the assumption that information is perfect”

The story, most likely apocryphal, I heard here is this: When he was working on the asymmetric information stuff Stiglitz was Arrow’s student. Now, Arrow, being a heavily left leaning sort, had been kicking himself for proving the First Theorem for awhile (apparently he originally thought that the technical conditions needed would be much more stringent). So when his student showed that relaxing the perfect information assumption invalidated the theorem he was pretty excited. In fact, he was so excited that when Stiglitz was presenting the paper in a seminar, Arrow jumped up, snatched the chalk away from him and presented Stiglitz’s paper for him.

”Stiglitz [says] that "[Economics as taught] in America's graduate schools... bears testimony to a triumph of idology over science." ... According to the authors:
"Economics courses at the undergraduate level typically place little-to-no emphasis on learning the tools of economic science, instead focusing on teaching algebraic simplifications of actual economic work, and then assigning problem sets in which students plug in values for the different variables””

Wait, so are we talking “the way economics is taught” (another much overused phrase) at the grad or undergrad level? Anyway, having just finished my own grad level indoctrination in the subject matter I’d say Stiglitz and those authors are pretty wrong on both counts (and the article is mostly argument by assertion)(Stiglitz is a great researcher but when it comes to commenting on “the state of the discipline” (AMOP) he has a tendency to say a lot of silly things – but of the kind that appeal to the prejudices of certain people):

In my experience of econ in grad school I hardly saw any ideology. We (i.e. grad students) even had a game among ourselves in which we’d try to guess the political persuasion of our profs – and in general it was almost impossible; hell I still don’t know what kind of politics most of them have. I remember one Political Science guy who was taking classes with us complaining that the Prof was “too free market oriented” being totally shocked when he went to office hours and found the shelves stacked with books on Socialism (this particular Prof was an economist in the old Marxian/Paretian tradition of Oskar Lange and Abba Lerner). You’re definitely NOT taught “free-market orthodoxy” or exclusively “neoclassical economics” – much of the time in fact is spent on analyzing market failures or newer theories like Behavioral stuff. The closest to it is when you do dynamic macro but even there you move really quickly from Prescott style “business cycle is the optimal adjustment to random supply shocks” to dynamic models with sticky prices, externalities, monopolistic features, or other non-Walrasian features. You don’t see much linear programming or neo-Ricardian stuff but… well, that’s a different discussion. I would agree/argue that there isn’t enough History of Economic thought, to put everything you learn in proper perspective, but then again I like that sort of stuff so maybe I’m just being biased – there’s a finite amount of time and a lot of stuff to squeeze into it.

My experience with “the way economics is taught” at the undergrad level comes from the other side, i.e. with actually teaching it.

Principles tends to be pretty dumb down, which should mostly be blamed on lack of any preparation prior to college.

Macro is pretty much still the old Keynsian IS/LM. You’re not gonna see RBC or Rational Expectations even at the intermediate level, except maybe for a brief mention that such theories exist somewhere out there (Dornbusch and Fischer pretty much do this). I guess closest to it would be Barro’s textbook and even there it’s sort of faked (and the textbook is rarely used, except maybe in some Masters programs). Linear programming theories might get a similar brief acknowledgment of existence. A good bit of time is spent on teaching basic definitions of GDP, unemployment, inflation, federal funds rate etc. That is “real world” stuff that will make the students able to understand the newspaper. You always spend a good deal of time on sticky prices and wages including usually the idea of efficiency wages (which is part of the research Stiglitz contributed to a lot). Anyway, I think this is mostly as it should be.

Micro is generally the old Marshallian stuff. Supply and demand. Partial equilibrium version of First Theorem, then rest of the course is pretty much why it often fails in practice. Externalities, taxes, monopoly, oligopoly. Then usually some application based on the Prof’s particular area of expertise. You might see some GE in an intermediate class. Constrained maximization might also make an appearance at the intermediate level but even there it’s generally presented as a tool not as an end in itself (contra the above quoted article). Again, I think this is mostly as it should be.

In arguments like these you always gotta ask “what are the alternatives?” The reason you don’t see RBC, GE, Game Theory, Linear and Dynamic Programming at the undergrad level is because these are pretty technical subjects – stuff that’s hard to explain with just a graph. And anyway I guess these are supposed to be the very “imaginary worlds” them French students are complaining about (which to me just says that imagination is exactly what they lack). IS/LM and Marshall are the main course of undergrad econ classes because they are simple enough to teach and simultaneously contain a “core” of what economics is about (even RBC style models these days look like dynamic IS/LM).

Basically the strategy of the article seems to be "throw a lot of imprecisely stated complaints" at "the way economics is taught" and hope something sticks.

But really, a lot of this “too much ideology in the way economics is taught” noise is really “NOT ENOUGH ideology in the way economics is taught”. Not enough MY ideology that is.

Anonymous said...

"In arguments like these you always gotta ask 'what are the alternatives?'"

Pretty typical defense, and one that doesn't show much imagination (good thing that the French students don't have a monopoly on that trait). The Post-Keynesians would say teach more Post-Keynesian theory, the radicals would say more radical theory, etc.

Since there are alternative approaches, with alternative research programs and so forth, why aren't these commonly taught in economics cirriculums? Presumably the "mainstream" school is superior in some way. However, the typical undergraduate microeconomics textbook does little to establish the positive successes of the discipline. Rather a methodology is presented, with an implicit promise that it is useful for understanding economic phenomena. A number of objections to methodology are completely ignored (like the Capital Controversies). The justification again typically ends up being your "what are the alternatives"? However, since the student is never alerted to these issues, there is no reason to familiarize oneself with any material beyond the canon.

A more pressing question: if economics is free of ideology, what accounts for the prominence of normative theory (especially in undergraduate teaching)? Unregulated markets produce too much pollution, we are told. By which criteria? The efficiency criteria. Is this a reasonable criteria to assess outcomes? And in particular, is it compatible with other ethical systems? The
use of this criteria becomes a convention of the discipline, but why? How can a school of thought claim to be scientific and yet yield so many normative trappings?