Agent-based Enterprise Competition Simulation Models For Industrial Cluster Study

Industrial Clusters have important effects on regional development whose sustainable development has received much attention by government. Cluster policy can restrain negative effect of industrial clusters and help improving suistainable development of industrial clusters as well. The starting point of our research is how to provide decision support for government to their cluster policy and how to examine the effects of different strategies of industrial clusters sustainable development.By reference to the theory of complex adaptive system, an industrial cluster is viewed as a socio-economic system including enterprise, government, university, research institutions, financial organization and industry association, etc. Among them, micro-enterprise is the most important entity in cluster. The current status of industrail cluster is the result of interactive influence and action among enterprises and between enterprises with cluster enviroment. Enterprises produce, sell and make technology innovation decision in industrial cluster. So cluster policy designer should consider how an enterprise makes decision in cluster. By having an influence on decision behavior of enterprises, the macro status of industrial cluster is changed indirectly and the sustainable development of industial cluster will enhanced. We also believe the idea of building a simulation model of industrial clusters on computer and then testifying the influence of different factor on industrial clusters a good way to give support for cluster policy design.Agent Based Modeling, one of the most interesting and advanced approaches for simulating a complex system, has been used to analyze industrial cluster. Though it is widely used in supply chain, economic system and social simulation, there is currently a lack of a unifying framework for agent based research. Based on many literatures about agent application in information network, software engineering, distribute computation, intelligent control and etc, a conceptual model of artificial society is put forward and adaptive behaviors are distinguished as high-level cognition and low-level adaptation. A summary of research methdology for socio-economic system research is given, in the light of comparative study between agent-based modeling and other traditional modeling. Such aspects in agent based economic and social simulation methodology, as the modeling opinion, life cycle, adaptive algorithm, applicability and universal programme architecture, are improved.By analysis of the development of Guangdong industrial cluster, current status of cluster policy and different entitis behaviors, a cause-effect model of sustaible deveopment of industrial cluster is built. By this model we conclude that capital accumulation and technology innovation are the two most important factors which affect the development of industrail cluster. According to the difference of competition between traditional industrial cluster and high technology industrial cluster, a Production and Marketing Simulation Model(PMS) and a Technology Innovation Simulation Model(TIS) are built by using agent-based economic and social simulation methdology after cases study on micro enterprise behaviors in cluster. By running the PMS model, we have observed the effect of the change of market demand, government tax, price and rent of realstate, market risk, labour market, industry profit, industry threshold, and finance factor on the capitcal accumulation of industrial cluster. The effect is positive, negative and U type in the experiments. Similarly, the effect of the change of research success rate, research investment ratio/rate, research cycle, return rate, market risk, goverment tax, socail network on the technology innovation of industrial cluster can also be observed by running the PMS model.Finally, we propose cluster policy should make clear the role of goverment in industrial cluster and encourage different innovation strategies in traditional industrial cluster and high technology industry cluster. We also bring forward that cluster policy design should consider cluster life cycle and take enterprise decision behavior into acount. Besides, cluster policy should prevent the gamble behaviors unfavourable to innovation and investment in industry cluster. The evaluation and statistics of indutry cluster should be encouraged. Seven solutions resulting from simulation are given to improve the sustainable development of industry cluster. The research findings provide opinions for cluster policy design and the modeling of economic and social system.