// Agent-based Simulations
In a connected and increasingly complex world it is often the case that simple models are no longer sufficient to answer research questions. The approach known as agent-based modelling was therefore developed as a new type of science involving computer-aided experiments for the analysis of complex problems. This can be used to set up an in silico policy laboratory, for example, in order to model the diffusion of innovations. It will at the same time allow an ex-ante examination of policies, such as funding measures, as part of the simulation and an appraisal of their impact before they are actually put in place.
The basis for the simulation is a flexible model developed at the ZSW which can be adapted to the research context in any given case. One example is the EMOSIM model which focuses on the analysis of diffusion in the field of electromobility. In this model, agents (here: households) interact with each other and make decisions on their choice of vehicle. They can choose between conventional vehicles (diesel, petrol), plug-in hybrids (PHEVs) and pure electric vehicles (BEVs). The decision will depend on various factors including the socio-economic profile of the households, their individual preferences in respect of the vehicle features, their affinity for innovation, the structure of the individual social network and its social norms. Some of these variables, e.g. socio-economic status, are assumed to be invariant over time while others are modified through the interaction of the agents in social networks during a simulation run.
