Agent-Based Modeling

Since the advent of computers, the natural and engineering sciences have enormously progressed. Computer simulations allow one to understand interactions of physical particles and make sense of astronomical observations, to describe many chemical properties ab initio, and to design energy-efficient aircrafts and safer cars. Today, the use of computational devices is pervasive. Offices, administrations, financial trading, economic exchange, the control of infrastructure networks, and a large share of our communication would not be conceivable without the use of computers anymore. Hence, it would be very surprising, if computers could not make a contribution to a better understanding of social and economic systems. While relevant also for the statistical analysis of data and data-driven efforts to reveal patterns of human interaction, we will focus here on the prospects of computer simulation of social and economic systems. More specifically, we will discuss the techniques of agent-based modeling (ABM) and multi-agent simulation (MAS), including the challenges, perspectives and limitations of the approach. In doing so, we will discuss a number of issues, which have not been covered by the excellent books and review papers available so far . In particular, we will describe the different steps belonging to a thorough agent-based simulation study, and try to explain, how to do them right from a scientific perspective. To some extent, computer simulation can be seen as experimental technique for hypothesis testing and scenario analysis, which can be used complementary and in combination with experiments in real-life, the lab or the Web.