Re: Are there theorems for Agent Based Models

[Benedikt Stefansson]

Paul Johnson wrote:
> 
> An economist said to me the other day, "where are your theorems? Aren't
> there established results that guide modeling choices?"  I shrugged.
> 

One of the best responses to these kind of statements that I've seen,
written by a noted mathematical economist, is Ken Judd's working paper: 
"Computational Economics and Economic Theory: Substitutes or
Complements?". His website has both PostScript and PDF versions of the
paper:

http://bucky.stanford.edu/

Highly recommended.

Judd points out that economists, unlike other social scientists or even
physicists for that matter, are obsessed with proving theorems. This is
supposed to make the field more "scientific", providing a firmer
foundation. Yet very few papers in economics actually build on theorems
proven in other economics papers. Each paper erects a new pristine piece
of scaffolding, which is then useless in other contexts because the
particular mathematical formalism which allowed the author do derive a
proof, was chosen in order to prove the theorem, not to fit the problem
that the model is meant to describe. 

Strangely enough, while theorem proving in the natural sciences is often
an 'exercise left to the reader' and does not a career make (Judd
maintains that Einstein for example hardly proved a single theorem about
the theory of general relativity) in economics the person providing
proofs for basic theorems usually becomes more famous than the person
which came up with the idea (e.g. the contributions of Debreu, Arrow,
McKenzie to general equilibrium theory). 

And in terms of 'that's good and well, but which established results
guide modeling choices in computational economics?' see the discussion
in the paper.

-Benedikt



-- 
---
Present coordinates: 
Dep. of Economics, Univ. of Trento, Via Inama 1, 38100 Trento, ITALY
Off: +39 0461 882246/267875 Mob: +39 347 0415721 Fax: +39 0461 882222


                  ==================================
   Swarm-Modelling is for discussion of Simulation and Modelling techniques
   esp. using Swarm.  For list administration needs (esp. [un]subscribing),
   please send a message to <address@hidden> with "help" in the
   body of the message.
                  ==================================