Empirical Calibration of Simulation Models

Simulations have become a common tool to study the implications of theoretical models. In addition, computational approaches are frequently used in statistics to adapt models to reality. We aim to merge these approaches. To obtain robust results with significant validity from simulations, empirical data have to be extensively used. We first discuss how inference in economic contexts can be made and then describe our proposed method. It makes extensive use of empirical knowledge for the development of a simulation model whose implications are then examined in the light of empirical data in a Bayesian-like approach. This allows all systems that can be described by the simulation model to be classified into subsets of models. This makes it possible to establish different subsets of models that describe certain realities and to study the characteristics of and causal relationships in these subsets