Dynamic microsimulation: general principles and examples in R

Dynamic microsimulation allows modelling complex systems where the variability of individual characteristics and behaviours plays a dominant role. Its basic principle is extremely simple but its implementation raises some methodological issues that are not always documented in descriptions of existing models. In addition, a major question for model builders is the choice of a programming language. Some dedicated platforms now exist. The simplicity of the method also allows starting from scratch with generalist languages and this is the choice that had been retained for the two successive versions of the Destinie model developed at Insee since the mid 1990s. A third option is to rely on a statistical package. This document provides a general introduction to this modelling technique and to its few methodological difficulties, and some examples of this third programming style, written in R, including a partial adaptation, in R, of the core of Destinie’s demographic module.