Multilateral attractiveness, migration networks, and destination choices of international migrants to the Madrid metropolitan area

Abstract

We investigate the effects of local characteristics on destination choices of international migrants to the Madrid metropolitan area. It is argued that the choice of a particular location does not only depend on the attractiveness of that location but also on the attractiveness of all other locations in the choice set. This is likely to be more important with a narrowly defined choice set with a fairly high degree of substitutability between functionally similar locations. We use the term “multilateral attractiveness” to refer to the unobserved influence exerted by all locations in the spatial choice set. Taking advantage of the equivalence relation that exists between Conditional Logit and Poisson, we estimate a location-choice model using the Poisson pseudo maximum likelihood estimator. Location-specific effects have been incorporated to account for unobserved spatial similarities (i.e., for possible violations of the IIA assumption). The fixed-effects Poisson model is applied to municipality-level migration data, for a set of five broadly defined groups of immigrants. The proposed estimation strategy has important empirical implications, where the magnitudes and/or signs of the estimated coefficients change in the expected direction. We find that the spatial pattern of immigrants’ location choices is fairly persistent over the time span considered. We also find that the impacts of local ethnic communities (network effects) are insignificant in some instances, while suggesting hetero-local settlement preferences or possible in-group rivalry in other instances.