If I understand the position correctly, both Chris Auld and I agree on
the following statements:
* Good fits of the Cobb-Douglas production function with the Solow
technique depend entirely and only on the constancy of relative
income shares.
* Solow says that this is a good result and what is to be expected if
you start by assuming the relevance of a production function and
marginal productivity relations.
* The production function and marginal productivity relations have
been shown formally to hold only in conditions that are remote from
any social system ever observed.
Unless Chris disagrees with any of the above points, everything else,
whether on his side or mine, is just rhetoric.
I am arguing that the second and third of the above points are examples
of a dysfunctional approach to social analysis. Chris, in common with
most economists including (if I understand her last email correctly)
Leigh Tesfatsion and the ACE community, believes that the second and
third of the above points are _not_ the result of a dysfunctional
approach to social analysis.
Chris noted that I hadn't dealt with his other objections -- my
discussion of the Radner theorem and Sargent's macroeconomics and
bounded rationality approach. Since it brings together Chris's and
Leigh Tesfatsion's objections, I will address the Radner theorem now.
Much the same points apply to the Sargent-inspired literature.
Consider the following two properties of all general equilibrium systems
with trading over time:
* All agents live forever.
* All agents have unlimited cognitive capacities.
>From Leigh's special issue introductions, her editorial decisions and
her contribution to this discussion, I infer that she believes we should
"look for the wheat" in systems based on such agent specifications.
Chris was more explicit:
> In fact, economics, like all scientific disciplines, uses
> models which simplify reality. Making rather implausible
> assumptions about computational abilities is just one such
> assumption. In this case, relaxing that assumption forms
> both a research agenda looking into its consequences and is
> often an issue in various models.
>
If we restrict ourselves to the modern formulations of general
equilibrium theory starting with Arrow and Debreu nearly 50 years ago,
it must be said that the relaxation of these assumptions is a long time
coming.
What makes economics dysfunctional is not that its models simplfy
reality. The dysfunctionality is a result of the acceptability and wide
reliance by economists on models demonstrably have no relationship to
reality whatever. (The Solow papers and the Kiyotaki-Wright paper cited
with approval by Tesfatsion are good but hardly unique examples.) A
point of some interest to the social simulation community is the
validation of representational models by which we mean the demonstration
that models usefully correspond to their target systems. This is
particularly important when the models are to be used for real system or
software design or to inform policy analysis.
Leigh claimed that ACE is a broad church and, on the evidence she
offers, that is clearly so. But also on the evidence offered, the ACE
community does not appear to be much concerned with model validation
with respect to real (as distinct from laboratory) social systems.
--
Professor Scott Moss
Director
Centre for Policy Modelling
Manchester Metropolitan University
Aytoun Building
Manchester M1 3GH
UNITED KINGDOM
telephone: +44 (0)161 247 3886
fax: +44 (0)161 247 6802
http://www.cpm.mmu.ac.uk/~scott
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|