Jan Pablo
Burgard studied economics at Trier University and made his PhD in statistics
covering topics in survey statistics and small area estimation. He habilitated
in statistics and econometrics and is now visiting professor at the Freie
Universität in Berlin. His main research directions are survey statistics,
applied statistics and statistical learning. One methodological framework where
all of these fields are necessary is microsimulations -- the topic of today's
talk.
Spatial dynamic microsimulations allow for the multivariate
analysis of complex systems with geographic segmentation. A synthetic replica
of the system is stochastically projected into future periods using micro-level
transition probabilities. These should accurately represent the dynamics of the
system to allow for reliable simulation outcomes. In practice, transition
probabilities are unknown and must be estimated from suitable survey data. This
can be challenging when the dynamics vary locally. Survey data often lacks in
regional detail due to confidentiality restrictions and limited sampling
resources. In that case, transition probability estimates may misrepresent
regional dynamics due to insufficient local observations and coverage problems.
The simulation process subsequently fails to provide an authentic evolution of
the system. A constrained maximum likelihood approach for probability alignment
to solve these issues is proposed. It accounts for regional heterogeneity in
transition dynamics through the consideration of external benchmarks from
administrative records. It is proven that the method is consistent. A
parametric bootstrap for uncertainty estimation is presented. Simulation
experiments are conducted to compare the approach with an existing method for
probability alignment. Furthermore, an empirical application to labor force
estimation based on the German Microcensus is provided.