Abstract

We examine the local determinants of destination choices of foreign immigrants to the Madrid metro area using data for 2005 and 2009 from the Spanish annual municipal-level registers of inhabitants. Taking advantage of the equivalence relation between conditional logit and Poisson, we estimate a location-choice model using the Poisson fixed-effects estimator. Origin-destination fixed effects are incorporated to account for the persistent spatial structure of the immigrants' settlement patterns and to control for potential violations of the independence of irrelevant alternatives (IIA) assumption. The Poisson regression model is estimated for seven different groups of immigrants according to world regions or countries of origin. Our modelling strategy has important empirical implications, with magnitudes and/or signs of the estimated coefficients changing in the expected directions. It is found that newly arriving immigrants tend to settle in low-to-middle-income locations in the suburban reaches of the Madrid metro area. Moreover, the effects of the size of local communities of established immigrants are found to be insignificant and even negative in several instances, reflecting hetero-local settlement preferences and/or the saturation of local networks causing in-group job rivalry, respectively.