In search of a modelling strategy for projecting internal migration in European countries - Demographic versus economic-geographical approaches

Internal migration is the most volatile and difficult to predict component of regional demographic change. A pure demographic approach using age and sex-specific parameters of migration intensities cannot fully capture the migration trends over time. One of the approaches that can be used for a better description of past trends and forecasting of future trends is to use additional non-demographic information such as regional economic indicators. In this paper we compare the predictive performance of pure demographic and extended economic-geographical models using data of four European countries at the so-called NUTS 2 level. The models are nested within a GLM specification%2C that allows both demographic and extended models to be written as specific cases of loglinear models. Therefore model fit and performance can be compared directly.