Forecasting Swiss immigration: a spatial dynamic panel data model
Alice Milivinti, Université de Genève
The rising importance of migration in the last thirty years is one of the components that made the world a co-integrated system. In a globalized economy the propagation of shocks, the convergence of regional incomes, the development of local policies and the contagion of financial crises are all relevant and interdependent phenomena that need to be taken into account for studying worldwide mobility. In this new context the complexity of describing the dynamics of the economics-population relationship has increased considerably. However, even if there are researches that try to model this new interdependent reality introducing cross-countries heterogeneity, virtually no study considers the interplay of different national characteristics in a global system. To the contrary, this paper estimates and forecasts Swiss immigration constructing a cross-country dataset where a dynamic spatial panel specification allows to analyse migratory flows resulting from resulting from path-dependent processes in which the geographical location is a significant part of the resulting equilibria. The estimated parameters generate out-of-sample predictions that display lower forecast errors than the usual alternatives. Starting from this empirical result the model is used to forecast Swiss immigration for the next twenty years.
Presented in Session 74: New directions in migration measurement