Multi-Agent Approaches in Economics

In social sciences and economics out of the mainstream, reality is Often understood as a complex system characterized - among others - by distributed interactions among heterogeneous agents. A somewhat different approach is in the mainstream economics, where representative agents - firms or households - are usually used. The advantage of the latter is that quite a lot of important results can be obtained analytically. If we work with heterogeneous agents, obtaining of analytical results is usually too complicated and computer simulation must be used. The aim of this paper is: to show that multi-agent approaches have been part of economic thinking for a long time, to illustrate the multi-agent approach in economics with some applications. From the well-known economists, the complex - multi-agent approach is clearly visible in the work of Friedrich A. Hayek and Herbert A. Simon. Their multi-agent thinking is illustrated in the first part of this paper. In the second part of this paper, the possibilities of the multi-agent approach in economics are shortly demonstrated on four applications. Applications concern the analyses of migration and migration networks and the analyses of the system of universities both with optimizing agents and agents with procedural rationality. The specific multi-agent model and simulation experiments for each application are described. The theoretical background is shortly summarized wherever necessary: motivation theory of A.H. Maslow and a simple neural network - perceptron - for the analyses of migration and migration networks and the difference between optimization and procedural rationality in the analyses of the system of universities.