Abstract

Experimental research is nowadays evidently the ‘gold standard’ of tax research. Simulations, especially with the so-called ‘agentbased models,’ are commonly used in economics as a research tool. Based on theoretical models, empirical results, and insights gained from experiments, simulations enable the study of whole societies structured by networks of agents, with respect to collective rather than individual reactions to taxation. The quality of the results, however, depends among other things on the inputs from empirical and experimental research employed to calibrate simulation models, as well as the structure of agent networks. Although simulations cannot in fact provide better explanations of tax compliance and evasion than economic and psychological research, they are nevertheless a new tool of research that yields additional insight into the aggregate dynamic behavior of agents in tax policy regimes.