Over-coverage in Swedish Population Registers Leads to Bias in Demographic Estimates
preprintposted on 03.09.2018, 08:36 by Andrea Fuentes Monti, Sven Drefahl, Eleonora Mussino, Juho Härkönen
Abstract: Estimating the number of individuals that live in a country has always been among the essential tasks for demographers. In this study we assess the potential bias in estimating the size of different migrant populations due to over-coverage in a country’s population-register system. Over-coverage, i.e., from individuals registered but not living in a country, constitutes an increasingly pressing phenomenon and is tightly linked to differential patterns of registered emigration. However, there is no common understanding on how to deal with over-coverage in demographic estimates and research. This study examines different approaches to over-coverage estimation and discusses ways of improving current estimation methods using Swedish total population register data for the years 1990-2012. We assess over-coverage levels across migrant groups, test how estimates of age-specific fertility and mortality are affected when adjusting for over-coverage, and examine whether over-coverage can explain parts of the healthy migrant paradox. Our results confirm the existence of over-coverage. When adjusting for over-coverage, we find substantial changes in mortality and fertility rates for people in migrating ages. Our results suggest that accounting for over-coverage is particularly essential for correctly estimating fertility in migrant populations.