Industry And Labor Dynamics: The Agent-based Computational Economics Approach, Proceedings Of The

431 Pages·2004·19.46 MB·English

by  Roberto Leombruni

industry and
labor dynamics
the agent-based computational
economics approach
TThhiiss ppaaggee iinntteennttiioonnaallllyy lleefftt bbllaannkk
edited by
Roberto Leombruni
Matteo Richiardi
LABORatorio R. Revelli, Italy
.m. il.s
industry and
labor dynamics
the agent-based computational
economics approach
proceedings of the
wild@ace2003w orkshop
Torino, Italy 3 - 4 October 2003
r p
World Scientific
-
NEW JERSEY * LONDON * SINGAPORE BElJlNG * SHANGHAI 0 HONG KONG TAIPEI CHENNAI
Published by
World Scientific Publishing Co. Re. Ltd.
5 Toh Tuck Link, Singapore 596224
USA once: 27 Warren Street, Suite 401-402, Hackensack, NJ 07601
UK once: 57 Shelton Street, Covent Garden, London WC2H 9HE
British Library Cataloguing-in-PublicationD ata
A catalogue record for this book is available from the British Library.
INDUSTRY AND LABOR DYNAMICS
The Agent-Based Computational Economics Approach
Proceedings of the Wild@Ace 2003 Conference
Copyright 0 2004 by World Scientific Publishing Co. Re. Ltd.
All rights reserved. This book, or parts thereoJ may not be reproduced in any form or by any means,
electronic or mechanical, includingphotocopying, recording or any information storage and retrieval
system now known or to be invented, without written permission from the Publisher.
For photocopying of material in this volume, please pay a copying fee through the Copyright
Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, USA. In this case permission to
photocopy is not required from the publisher.
ISBN 981-256-100-5
Printed in Singapore by World Scientific Printers (S) Pte Ltd
Contents
The Wild@Ace Project vii
B. Contini, R. Leombruni and M. Richiardi
Section 1 Methodology 1
Economics and Complexity 3
A. Kirman
Simulations, Theory and Experiments. Notes from an 22
Historical Perspective
M.N ovarese
JAS: Java Agent-Based Simulation Library, an Open Framework 43
for Algorithm-Intensive Simulations
M. Sonnessa
Section 2 Microsimulation of Labor Dynamics 57
Matching, Bargaining, and Wage Setting in an Evolutionary 59
Model of Labor Market and Output Dynamics
G. Fagiolo, G. Dosi and R. Gabriele
Endogenous Matching Functions: An Agent-Based Computational 90
Approach
M. Neugart
A Search Model of Unemployment and Firm Dynamics 107
M. Richiardi
Evolution of Worker-Employer Networks and Behaviors Under 129
Alternative Non-Employment Benefits: An Agent-Based
Computational Study
M. Pingle and L. Tesfatsion
Early Retirement from the Labor Market: Policy Experiments in 164
the Italian Case
Vagliasindi, M. Rornanelli and C. Bianchi
V
vi
Section 3 Understanding Firm Behavior 197
Business Cycle Fluctuations and Firms’ Size Distribution Dynamics 199
D. Delli Gatti, C. Di Guilmi, E. Gaffeo, G. Giulioni,
M. Gallegati and A. Palestrini
Effects of the Interaction of Heterogeneous Rationalities on the 222
Innovation Output of Firms -A Multi-Agent-System Approach
A. Kaufmann
Modular Pyramidal Hierarchies and Social Norms. An Agent-Based Model 244
A. Dal Forno and U. Merlone
Section 4 Industrial Clusters and Firm Interaction 257
Growing Silicon Valley on a Landscape: An Agent-Based Approach 259
to High-Tech Industrial Clusters
J. Zhang
Simulating Knowledge Dynamics in Innovation Networks 284
€? Ahrweilel; A. Pyka and N. Gilbert
A Generalised Computational Model of Firms Production and 297
Interactions: Preliminary Results on Industrial Dynamics
7: Ciarli and M. Valente
Labor Market, Entrepreneurship and Human Capital in Industrial 332
Districts. An Agent-Based Prototype
R. Boero, M. Castellani and E Squazzoni
The Role of Small Business Based Structures in Promoting Innovation, 350
and Creating Employment, as Compared to Oligopolistic
Based Structures
M. Khoshyaran
Section 5 Mathematical Tools 375
A Finitary Approach to Clustering 377
D. Costantini, U. Garibaldi and €? Viarengo
Conclusions 399
Open Problems in Using Agent-Based Models in Industrial and 40 1
Labor Dynamics
N. Gilbert
THE WILDQACE PROJECT*
BRUNO CONTINI
Universith di Torino t& LABORatorio Revelli %
bruno. [email protected]
ROBERTO LEOMBRUNI
LABORatorio Revelli
[email protected] t
MATTE0 RICHIARDI
LABORatorio Revelli
m. richiardi@labor-torino. it
Keywords: Computational economics; agent-based simulation; economic
methodology.
1. The Wild@Ace Project
From time to time, innovative methodologies are introduced to the mod-
eling arena which produce a sense of great excitement in some researchers
and a sense of great irritation in others. The latest of these self-proclaimed
“revolutions” may very well be Computational Economics, among whose
enthusiast non-practitioners stands Richard Freeman. In his well-known
‘‘War of the Models” article 5, Freeman expresses great faith in the po-
tential of several techniques which lie at the intersection of Evolutionary
Economics, Computer Science and Cognitive Science:
*the authors are grateful to the participants of the wild@ace 2003 conference for their
valuable comments and suggestions. usual disclaims apply.
tuniversitk di torino, dipartimento di economia, via PO 53, 10124 torino, italy
laboratorio revelli, via real collegio 30, 10024 moncalieri (torino), italy,
http://www.labor-torino.it
vi i
viii
Our empirical tools are wonderful for ceteris paribus problems,
but many issues regarding labor institutions are mutatis mutandis
problems. Lots of interrelated changes with no empirical counter-
factuals. This implies that if we are to make progress, we need
something more in our tool bag. Game theory? A language and
framework, but not sufficiently specific. General equilibrium? Too
[...I
general and static. Then what? There are a new set of
theoretic and empirical tools that seem suited for the problem of
analyzing labor systems and the War of the Models. The tools
range from theoretical simulations of nonlinear dynamic systems
to a theoretic data-mining. Complexity analysis. Neural networks.
Data-mining for knowledge discovery. Landscape models. Artificial
agent simulated societies. Chaos theory. Complex adaptive sys-
tems. Nonparametric statistical tools of diverse shapes and sizes.
Cellular automata. The hills are alive with the sound of new tools
and jargon.
True, simulations have been around in Economics for years, although
they have been used mainly for predictive purposes. And, true, simulations
have played an important role in only a limited number of relevant results,
whether these be theoretical or empirical. In fact, one often has to draw on
the classic work of Schelling 16, in order to defend the simulation approach
against the arguments of skeptical economists. Yet despite all this, and
perhaps unbeknownst to the general public, a vigorous strand of research
is growing and developing in the field. The Wild@Ace project forms part
of this movement. The acronym, which stands for “Workshop on Indus-
trial and Labor Dynamics The Agent-based Computational Approach” ,
-
identifies the intersection between a specific area of interest , namely Labor
Economics and Industrial Organization, and a methodology, the agent-
based simulation approach. The project was inaugurated in 2003, with the
first Wild@Ace conference held on October 3-4 at the LABORatorio Rev-
elli Centre for Employment Studies in Turin, Italy. The workshop, which
was jointly organised by the LABORatorio Revelli, the Department of Eco-
nomics “S. Cognetti De Martiis” of the University of Turin, and the Social
Interaction Economics and Computing group (SIEC) of Ancona (Italy) ,
was the first event ever to have this particular focus. The workshop will be
run annually.a This volume is a collection of the papers presented at the
aThe next workshop is scheduled to be held again at the LABORatorio Revelli in De-
cember 2004.
ix
Wild@Ace 2003 conference.
This introduction is structured as follows. In section 2 we briefly dis-
cuss the meaning of agent-based simulations and their potential for use in
economic modeling. Then, in section 3 we elaborate on fieeman’s hot list
of topics for which traditional (algebraic) modeling has proved to his and
-
our judgment insufficient, focusing in particular on the role of institutions.
-
Then, we turn to the appropriateness of agent-based simulations for taking
on these challenges (section 4). In particular, we respond to the skepticism
that mainstream economists often express toward this novel methodology.
We analyze the main objections that can be made against agent-based sim-
ulations, and show that appropriate solutions do exist. Finally, Section 5
summarizes the main contributions to this volume.
2. Why Write an Agent-based Simulation?
Agent-based models are computer models in which a multitude of agents
- each embodied in a specific software code - interacts. These agents can
represent individuals, households, firms, institutions, etc. Moreover, “spe-
cial” agents can be added to observe and monitor individual and collective
behavior.
Agent-based simulation models have two distinguishing features. One
is that they are agent-based, i.e. they follow a micro approach; the other is
that they are simulation models, i.e. they follow an inductive approach to
the discovery of regularities. Neither feature is exclusive to the methodol-
ogy. Many algebraic models have micro-foundations (e.g. Game Theory) ,
and many simulation models adopt an aggregate perspective (e.g. System
Dynamics). However, it is the intersection of the two approaches that de-
fines the methodology and that permits the extreme flexibility in design of
the model, while avoiding all the problems connected with merely aggre-
gate representations of the world ’, lo. Agent-based simulation models are
a third way between fully flexible but not computable and hardly testable
literary models (which provide no more than a verbal description of the
causal relationships behind a given phenomenon) on one side, and more
transparent but highly simplified algebraic models on the other side 6. The
biggest advantage of ACE models over the algebraic approach is their flex-
ibility, since the results are computed and need not be solved analytically.
With ACE, the researcher gains almost complete freedom over the specifi-
cation of the interaction structure and individual behavior.b
bAs is always the case, this freedom must be exercised with caution. With less of a need