Strategic interaction in oligopolistic markets — experimenting with real and artificial agents

Abstract

Distributed Artificial Intelligence (DAI) concepts and ideas are used to model dynamic strategic interaction among producers in oligopolistic markets. Results of a laboratory experiment with human agents in the role of duopolists were used to design an architecture of artificial agents. A system in which these artificially intelligent agents interact as duopolists in a market environment was implemented and tested. The artificial agents exhibit some of the learning and adaptive capabilities observed in human subjects and the system is able to reproduce the overall pattern of interaction of the laboratory experiments. Results so far encourage us to think that the implementation of a computational test-bench for research in economic strategic interaction based in DAI techniques is possible and may contribute to theoretic progress in economics.