Doktorarbeit

A continuous-time microsimulation and first steps towards a multi-level approach in demography

Zinn, S.
XXII, 222 pages. Rostock, Universität Rostock (2011)

Abstract

To face the challenges of aging societies, policy makers need reliable information about future population compositions. Microsimulation is a methodology that lends itself to this objective, as it very closely mimics life-course dynamics. In this thesis, we describe the development of the demographic microsimulation with a continuous time scale that we have realized in the context of the project MicMac - Bridging the micro-macro gap in population forecasting. Furthermore, we detail extensions that we have added to the initial version of the MicMac microsimulation. To avoid any redevelopments and to make use of up-to-date modelling and simulation (M&S) technology, we used the DEVS formalism to specify the microsimulation model, and we employed the M&S framework JAMES II to realize it. Due to its modular and state-based concept, the DEVS formalism is well suited to formulate population dynamics in the intended way. The usage of JAMES II allowed us to profit from already implemented M&S functionality and to validate the final microsimulation product. In a second step, to overcome the unrealistic assumptions of independent life-courses, we worked out a method to match individuals. In our example, we defined a match when individuals experienced the onset of a marriage or cohabitation. Here we tried to mimic human partner choice and to aim at assortative mating. Testing the mate-matching method shows that it produces feasible results that are in accordance with observed mating behavior. To model life-course dynamics after spouses are matched, we designed an approach that considers couples as entities and therefore describes the life-course dynamics of spouses jointly. We set up a ml-DEVS model to specify a microsimulation model that considers couple dynamics. The macro-DEVS model comprises two types of micro components: individuals and pairs. The micro components handle the life-course dynamics of unlinked individuals and the dynamics of married or cohabiting couples, while the macro-DEVS model guides the onset of partnerships (marriages or cohabitations), i.e., it performs mate matching and instructs the creation of pair models. We conducted a small case study to illustrate the capabilities of the microsimulation approach designed. Based on a synthetic population, we could show that it performs well.
Das Max-Planck-Institut für demografische Forschung (MPIDR) in Rostock ist eines der international führenden Zentren für Bevölkerungswissenschaft. Es gehört zur Max-Planck-Gesellschaft, einer der weltweit renommiertesten Forschungsgemeinschaften.