Agent-based Micro-simulation of Business Establishments

This paper describes the development and t esting of an agent-based micro-simulation of the evolution of individual 'business establishments' in a spatial economic system. The simulation uses comparatively simple, yet behavioural, rules and probabilistic models, using a Monte Carlo process to simulate behaviour from the probabilistic models. Establishments are represented by their consumption of commercial floorspace along with their production and consumption of other commodities. Markets for commercial floorspace are represented by spatially disaggregated "price landscapes", with the price at each point adjusted based on previous shortages or surpluses. Discrete choice models are used to represent establishments' choice of locations within this price landscape where they pursue their business of purchasing and selling commodities. These models provide a measure of the composite utility of selling or purchasing commodities from any business location, which are used to determine the degree of success of an establishment. This attractiveness measure is used to determine the probability of establishments growing, becoming bankrupt, leaving or splitting ("spin-off" or "duplication"). The system is fully integrated with a floorspace development simulation, which simulates the decisions of each landowner with regard to development, redevelopment or demolition. The price landscape (and vacancy landscape) for various floorspace types influences developer decisions. The test system is represented using a 10x10 system of zones. This test system is used to test the role of the various parameters, to determine reasonable values for the parameters, how the model behaves when parameter values are unreasonable, and how each parameter influences the model system.