Modelling kinship networks using analytical and microsimulation approaches

Analytical and microsimulation models are the primary approaches for estimating kinship networks when detailed, long-term demographic data including kinship ties are unavailable. Although both methods use population-level demographic data as inputs, their methodological formulations, assumptions and outputs differ. Analytical models offer mathematical transparency, enabling the calculation of expected values, but are limited in capturing individual variability and interactions. Conversely, microsimulation models simulate demographic events at the individual level, enabling the estimation of the number and distributions of kin, but they are computationally intensive and rely on additional assumptions. Despite their widespread use, systematic comparisons between these approaches remain limited. This study examines the consistency between matrix kinship models and Socsim demographic microsimulation when estimating kinship networks under comparable assumptions. Using fertility and mortality data from the World Population Prospects 2024 for Sweden (1950--2100), we compared the means and distributions of the numbers and ages of different kin estimated from both models. Our results show that when their main demographic assumptions are harmonised, both models produce similar estimates of the mean numbers and ages of kin, despite some fluctuation in the simulated estimates across ages. However, adding assumptions to the microsimulation, such as minimum birth intervals, alters this agreement. Keywords: Kinship, Demographic Microsimulation, Formal Demography of Kinship