Microsimulation

Microsimulation offers the tools needed to undertake analysis of individuals rather than on aggregated data in order to capture the heterogeneity and distributional effects of different assumptions on a population. Microsimulation covers a range of methods for the creation of synthetic datasets of individuals, the assessment of short-term changes which impact those individuals, and dynamic models which incorporate time and allow for the assessment of longer-term impacts of policy or other changes on the population. Where geography is introduced as an identifying feature for these individuals, the methods are known as spatial microsimulation. The increase in availability of microdata, from survey and other sources, combined with increases in computing power have meant that microsimulation methods have been adopted for a diverse range of research applications, and the field continues to evolve and grow. This chapter provides the background to microsimulation techniques and offers discussion of the practical considerations when building and validating models. An overview of deterministic and probabilistic methods used to create spatial microdata is provided and a discussion of validation techniques used to assess the quality of synthetic outputs is offered. Static microsimulation is discussed in the context of studies which have utilised the approach to assess short term effects of policy change and the extension to dynamic microsimulation is considered using a number of examples. The treatment of time in dynamic models and their extension to the spatial is covered. Throughout the chapter reference is made to literature which discusses key concepts in more detail and to examples where these models have been used in applied contexts. As a primer for the various methods which comprise microsimulation, the chapter provides an overview of the field, examples and practical considerations for implementing these models, and guidance on where to look for more detailed information on specific elements.