A simulation study of an idealised world of agents trading a single generic product is presented. The population consists of agents using forward planning and agents reacting to the current situation who are in direct interaction and competition. The object of the simulation is to study the relative efficacy of planning and reactive policies for agents in an environment that is complex, partly unpredictable and created by the actions of the agents themselves. In such a simulation, the policies can be compared in terms of policy robustness and policy risk as well as the traditional optimality. By including only the essential features of trading in the model it has been possible to thoroughly investigate the parameter space and draw rather general conclusions.
|Publication status||Published - 1 Apr 1996|
- Complex Systems
- Inventory Analysis