Can we use NEG models to predict migration flows? An example of CEE accession countries

In this paper we develop an analytically solvable and structurally estimable economic geography model and apply it to predict migration flows for the period following the CEE’s integration with the EU. The main innovation of our approach is that it endogenises both explanatory variables and the migration rate. The underlying structural parameters are estimated econometrically using a migration equation, which is derived entirely from the theoretical NEG model. Our simulations show that even relatively moderate changes in some of the explanatory variables (such as transport costs) can actuate unpredictable changes (both in sign and magnitude) in other explanatory variables (such as wages). Keeping these explanatory variables fixed, as in reduced-form models, would produce biased results. Our empirical findings advocate that there is enough evidence to predict a selective migration among the three Baltic States. However, labour mobility in the Baltic countries is sufficiently low to make the swift emergence of a core-periphery pattern very unlikely at this geographical level.