|
post-autistic
economics review |
|
issue 31
contents
PAE
Review index
home page |
|
Comment on McCartney Peter Dorman (Evergreen College, USA) ©
Copyright: Peter Dorman 2005 I agree with the general argument presented in McCartney (2005) regarding game theory and multiple equilibria. Neoclassical economics has, in its discomfort with this consequence of strategic interaction, attempted to suppress it through implausibly strong assumptions that began with Nash and continue to the present. Removing this straightjacket is a crucial first step toward recovering the intellectually liberating potential of game theory. I would like to make three further points, which I offer as “friendly amendments” to McCartney. 1. It is important to identify and analyze the assumptions that suppress multiplicity of equilibria. A wonderful example can be found in Hargreaves Heap and Varoufakis (1995). The central assumption in traditional neoclassical theory is the (quasi-)convexity of preference and feasible production sets, as I argued in Dorman (1997). 2. McCartney emphasizes the policy aspects of equilibrium selection. Once it is recognized how common (and multiple) such multiple equilibria are in complex economies, however, it is clear that even the most conscientious policy makers would be unable to exercise selection across all of them. In our own largely decentralized economies, the main role is played by history: equilibria are “selected” by virtue of their being in the neighborhood of past equilibria. This path dependence is ubiquitous and constitutes a major reason for the centrality of history in the study of economics. 3. The motivation of neoclassical theorists in suppressing multiplicity of equilibria raises important questions of scientific method. According to the dominant point of view, a theory must produce a unique prediction (identify a unique equilibrium) in order to generate and motivate hypotheses for empirical testing. It is clear that the unfettered game theory McCartney and I advocate does not do this. Is that a problem? It depends on the role that theory is called upon to play in relation to empirical work. The product of the neoclassical approach to theory-building is a set of categorical hypotheses. The components of such a theory, including the assumptions employed to impose unique solutions in games and other models, can be entirely arbitrary so long as logical consistency (especially with the postulate of rational choice) and empirical confirmation of the ensuing hypotheses are upheld. There are two main criticisms that can be made of this conception of “economic science”: a. In reality, hypotheses are rarely subjected to critical empirical tests. Rather, econometric work typically relies on a large number of supportive assumptions that can be adjusted as needed to permit a result “consistent with” the theoretical framework. This was the situation I encountered in my own work on the theory of compensating wage differentials, for example. No economist had sat down and asked, what assumptions and implications of this theory can be put to the test, and what sorts of tests would strip away as many confounding factors as possible? (Dorman, 1996 and Dorman and Hagstrom, 1998) The more general point is this: if a theory is to be assessed entirely on the basis of its consistency with empirical evidence, then an extra burden falls on empiricists to be as aggressive as possible in the construction of critical tests. Of course, it may be the case that no economic theory could survive under such an assault, which would mean that reliance on empirical testing to justify arbitrary theoretical assumptions (such as those that sustain unique equilibria) is not a defensible strategy. b. Even should a body of theory survive the most critical testing, however, its explanatory potential is minimal if, in fact, it does not explain. To explain means to provide a mechanism that accounts for the relationship between causal and consequential states. We need such mechanisms not only to satisfy our intellectual curiosity, but also to know when and how to deploy a theory. No mechanisms, save perhaps for a few in the most fundamental physical processes, are universal. To apply a theory normally means to identify situations in which the mechanisms it posits can be expected to operate. Neoclassical economics claims to have little interest in mechanisms, least of all in the most basic process in its chosen field of study, market equilibration. (Hence its black box character, particularly in the form of tâtonnement.) The alternative view of scientific work embraces the central role of mechanisms. This is how geology, biology, chemistry and virtually every other science (with the possible exception of physics) functions. In economics it means that theory would consist of an assortment of well-described social processes pertaining to the economy. Such processes might or might not have determinate outcomes, but the mechanisms they embody would provide a basis for context-specific empirical and policy analysis. This is not the place to make the case for such an approach, but it is appropriate to mention that game theory, in its open-ended, multiple-equilibrium conception—unencumbered by assumptions whose main purpose to impose a “solution”—is an ideal tool for elaborating mechanisms in complex social situations. References Dorman, Peter. 1997. Nonconvexity and Interaction in Economic Models. Social
Science Research Network Working Paper No. 1232, Microeconomic Theory 2(2),
2/12/97. Dorman, Peter. 1996. Markets and Mortality: Economics,
Dangerous Work and the Value of Human Life. Cambridge: Cambridge
University Press. Dorman, Peter and Paul Hagstrom. 1998. Wage Compensation for Dangerous Work
Revisited. Industrial and Labor Relations Review. (52)1: 116-35. Hargreaves Heap, Shaun and Yanis
Varoufakis. 1995. Game Theory: A Critical Introduction. London: Routledge. McCartney, Matthew. 2005. Game
Theory: A Refinement or an Alternative to Neoclassical Economics? post-autistic economics review. 30 (21
March): article 2. http://wwww.btinternet.com/~pae_news/review/issue30.htm ___________________________ |