Title: Small area estimation of transition probabilities for spatial dynamic microsimulation models in socio-economic research
Authors: Simon Schmaus - Trier University (Germany) [presenting]
Jan Pablo Burgard - Trier University (Germany)
Joscha Krause - Trier University (Germany)
Abstract: Dynamic microsimulations allow for the analysis of complex socio-economic systems. First, a synthetic replica of the corresponding system on a micro level as the base population is generated. Next, the replica is stochastically projected into future periods under different scenarios. Comparing the simulation outcomes then provides insights on essential system properties. The projection requires the definition of transition probabilities for every system-intrinsic entity. In order to obtain authentic simulation results, they must reflect the characteristic dynamics of the system. As corresponding probabilities are unknown in practice, they must be estimated from survey data. However, transition probability estimation can be challenging. If the system is spatially segmented into heterogeneous areas, dynamics may vary locally and regional estimation is required. Regional probability estimates may be subject to high uncertainty if the survey data lacks in local observations. We discuss methods of small area estimation and regional benchmarking for transition probability estimation in spatial dynamic microsimulations. The methodology is demonstrated in a simulation study based on a dynamic microsimulation model for care research in Germany.