| Abstract |
This research presents an alternative methodology that complements the forecasting of migration patterns. Tested across various regions of Spain using a theory-based migration time series, this approach offers a perspective that provides valuable insights into migration dynamics. To do so, we apply an analytical setting based on a parsimonious representation to project the migration time series of Spain through the reduction of the dimensionality of large sets of variables used in migration theory. This would allow us to verify whether the inclusion of multiple variables enhances the prediction of migratory flows by employing a structure based on dynamic factor models (DMF). Accordingly, the estimation accuracy is evaluated against the traditional multivariate representation of vector autoregressive (VAR) models. To conduct this assessment, the analysis is oriented towards migration from the most populated autonomous communities, i.e. Andalusia, Castile-Le & oacute;n, Catalonia, and Galicia, to Madrid. Thus, this paper seeks to confirm that the covariables recognised in theory for explaining internal migration within Spain, as a whole, serve as a reliable basis for forecasting future migration flows. |
