Tuesday, February 05, 2008

Problems Mount In Application of Free Market Theory

As a footnote in a discussion of anomalies in a paradigm, in the sense of Thomas Kuhn, Pasinetti writes:
"It seems to me significant that Kenneth Arrow, himself a major contributor to the clear formalisation of the present dominant paradigm based on the Walrasian exchange model of General Economic Equilibrium analysis, would write a short newspaper article (rather than an article in a scientific journal)... While giving a proud statement of the exchange paradigm, Ken Arrow frankly gives at the same time a long list of its major points of weaknesses and downright failures, implying that the number and the seriousness of these failures are continually mounting." -- Luigi L. Pasinetti, Keynes and the Cambridge Keynesians: A "Revolution in Economics" to be Accomplished, Cambridge University Press (2007)
Pasinetti refers to:
  • Kenneth Arrow, "Problems Mount in the Application of Free Market Theory", in the rubric "Debate" in The Guardian, 4 January 1994.
I could not find this article on-line by using search tools. I'm curious what Arrow could have written. Could he have explained the Sonnenschein-Debreu-Mantel results in that forum? The difficulty of developing something more realistic than the tâtonnement process?

Update (6 February 2008): I thank an anonymous commentator for posting Arrow's article in the comments.


Anonymous said...

Here it is...(retrieved from Nexis so no link)



The Guardian (London)
January 4, 1994


THE core of standard economic analysis and a major basis for its numerous applications in the world of policy is the theory of general competitive equilibrium (GCE).

The theory is competitive because it assumes that economic agents are predominantly small in the markets on which they operate - that is, the buying and selling decisions of any one agent do not seriously affect market prices.

It is an equilibrium theory in that supply and demand balance for all goods and services. It is general in that the markets are interrelated - a change in price in one market feeds through to other markets.

The theory emphasizes a complex set of interactions, so it is frequently simplified in application - but the underlying logic has been central in both pure analysis and policy formation.

The logic is implicit even in the earliest writings of mainstream economic theory, as in Adam Smith and David Ricardo. But it was given its modern formulation by the French economist Leon Walras and greatly extended to include the roles of time and uncertainty, by the late John Hicks, Maurice Allais, and Gerard Debreu and myself.

A basic assumption of competitive models has been that of constant or diminishing returns to scale in production - that is, it is impossible for a firm to cut unit costs by increasing output.

That many kinds of production exhibit increasing returns has been recognised intermittently from Adam Smith on, but has proved difficult to incorporate into the main body of theory. This is true despite developments in the theory of monopoly, oligopoly, and imperfect competition - currently synthesized under the elegant framework of game theory.

Concepts of equilibrium have proved harder to define. Indeed, the beauty of the competitive model is that different equilibrium notions lead to the same conclusions, but this equivalence no longer holds when markets are not competititive.

More serious, however, has been the inability of economists to construct a general theory incorporating increasing returns and imperfect competition.

GCE has served better than might be thought, despite so much abstraction from the real world. For one, protectionist trade policies based on overinterpreted theories of monopoly and increasing returns have proved inferior to free trade.

For another, rational expectations models based on GCE have proved considerably better than might be expected in explaining the path of share prices and even of cyclical fluctuations in the economy - though they are still far from adequate, as explained below. There are many other examples, for instance with regard to the effects of taxes.

But a long list of empirical failures mark the application of GCE. Some are literal falsifications of the model; some are omissions, important aspects of the economy which the theory does not address; and some are questions about the theory's presuppositions. I list only a few.

The best-known falsification is the recurrent and now chronic existence of mass unemployment, which is a straightforward contradiction of equilibrium. Closely related is the fact that businesses find that sales are constrained by the demanders, not by a company's willingness to produce at given prices.

Second is the excess volatility of share, oil, metal and other mineral prices, which are hard to explain as movements in expected discounted values of future prices. There is an even deeper difficulty with minerals prices. GCE and rational expectations imply steadily rising prices for exhaustible resources, yet even long-term trends have often proved remarkably flat.

As for share prices, there are a number of other difficulties in reconciling the actual course of prices with any form of rational expectations. In particular, the observed data on neither the excess return on equities compared with bonds nor the volume of trading on securities (and other financial markets) can be explained in these terms.

Two very different omissions in GCE are long-term growth in per capita income and the widespread existence of contracts which depart from the price system (eg, limits on insurance contracts, executives' incentive payments, rationing of bank credit).

GCE in a straightforward form has only capital accumulation as an explanation for growth, but this implies falling and eventually zero rates of per capita income growth.

It can be saved by assuming that increasing productive knowledge is available from outside the system ("exogenous growth"). But this is unsatisfactory intellectually and does not fit the real world.

In interesting but as yet not well-understood ways, acquisition of knowledge as an economic activity is a form of increasing returns, a point of view represented in the "new growth economics" of Paul Romer, Paul Krugman, and others.

But innovation is clearly outside the usual frame of GCE or even more general economic formulations. It is more nearly akin in some ways to biological evolution, with natural selection for successful ideas, and this idea has been broached by such students of innovation as Richard Nelson and Sidney Walter.

The understanding of complex non-price contracts is currently based on the idea that information differs from one economic agent to another, and compensation arrangements are drawn up with this in mind. This theory has developed remarkably in the past 30 years.

Two presuppositions of GCE that have been challenged are constant returns to scale in production and the excessive assumptions about the rationality of economic agents. As GCE has become more sophisticated, especially in the rational expectations version, the supposed rationality of agents has grown.

The sheer complexity of the computations implied in the theory already indicates the difficulty in assuming that individuals or firms can really be rational in the full sense.

There are doubtless deep connections between the limits of individual rationality and the properties of asset markets which seem aberrant from the GCE viewpoint.

The idea of economic agents striving to learn from a changing environment, with actions that are themselves part of that environment, fits naturally with the evolutionary viewpoint needed to study innovation and with modern ideas in artificial intelligence. Some economists, physicists and biologists associated with the Santa Fe Institute have been attempting to draw these ideas in a new synthesis, but it remains to the future to assess their success.

Kenneth Arrow is Professor Emeritus of Economics at Stanford University and a winner of the Nobel Prize for Economics in 1972.

Gabriel Mihalache said...

The title is far more sensationalist than the content.

I hope we'll agree that articles titled "Problems Mount In Application of Marxian Theory" and "Problems Mount In Application of Old Keynesian Theory" could, all in all, be far more damning, especially if we control for the number and variety of applications that each school gets.

Beyond that, I find that some of Arrow's comments are misleading:

Fama makes clear that the equity premium puzzle is a puzzle for consumption-based, representative agents asset pricing models. These are further strong restrictions on the GCE framework.

Handwaving about biological evolution? Marshall did too. Show me the money, I say.

You can have gains from international trade and growth with constant returns. Prescott shows us how.

Lastly, in what sense is Sonnenschein-Debreu-Mantel a problem?

YouNotSneaky! said...

Man, that article is soooooooooo neo-classical it hurts.