erc/metu
INTERNATIONAL CONFERENCE IN ECONOMICS  IV
September 13-16, 2000, Ankara

Agent-based Simulation of Economics

Jaroslav Zajac (Slovak University of Technology, Bratislava)

Abstract  

Agent - based simulation reports a project in agents-based computer simulation of processes of economic growth in a population of boundedly rational learning agents. The same family of growth models will be simulated under different assumptions about the nature of the learning process and details of the production and growth process. The purpose of this procedure is to establish a relationship between the assumptions and the simulation results. The simulation techniques with baseline simulations of boundedly rational learning processes that do not involve the complications of dealing with long-run economic growth. Can simulation "explain" the puzzles of economic growth and particularly the key puzzle of growth and learning processes that produce the puzzling results? And just what assumptions of the simulation are not predictable associated with puzzling results?

Many agent-based systems rely for their effectiveness on the individual agents, among interaction agents is required to coordinate these individually complex decision. An agent may take an effort a set of efforts and the output will depends on the random state of nature. The output will rise stochastically as the level of forts increases and the state of nature improves additional information is valuable because it allows a more accurate judgement of performance of the agent. Typical problems is multi-agents case can also offer opportunities for inference of hidden action and agent monitoring agent. Multi agents require a hierarchical organization t o coordinate tasks and monitoring agents´ input and information costs can be reduced and productivity improves.

Economic Research Center
Middle East Technical University
06531 Ankara Turkey
Phone: + 90 312 210 3044, 210 2003
Fax: +90 312 210 1244
e-mail: metuerc@metu.edu.tr