Adaptive learning in evolving task allocation networks

  • Authors:
  • Tomas Klos;Bart Nooteboom

  • Affiliations:
  • Delft University of Technology, Delft, The Netherlands;University of Tilburg, Tilburg, The Netherlands

  • Venue:
  • Proceedings of The 8th International Conference on Autonomous Agents and Multiagent Systems - Volume 1
  • Year:
  • 2009

Quantified Score

Hi-index 0.00

Visualization

Abstract

In this paper, we study multi-agent economic systems using a recent approach to economic modeling called Agent-based Computational Economics (ACE): the application of the Complex Adaptive Systems (CAS) paradigm to economics. In this paper, we apply the CAS paradigm to the study of an industrial goods market, where firms need to decide between making and buying components. Computer simulations using our model explain different emerging distributions of economic activity among organizational forms (market and hierarchy) in terms of the search problem facing the agents, and in terms of the negative consequences of the agents' search behavior on their perceived trustworthiness in the eyes of their potential partners. A further impediment to reaching optimal allocations we observe is that agents learn to protect themselves and their current allocation by being loyal and by focusing on their trust in their partner, rather than their partner's profit generating potential.