Inter-State Migration in Mexico, 1995-2000: A Flexible Poisson Gravity-Modeling Approach

Abstract

In this paper, I present an empirical study of aggregate inter-state migration in Mexico in the period 1995-2000. The study contributes to the empirical literature on migration in a number of ways. Firstly, I show that the gravity model in levels (Poisson model) provides a more realistic description of the migration process than a model in logs (log-linear model). Secondly, I develop a more flexible model by relaxing the standard assumption of spatial invariance (constancy) of the distance-decay parameter. This added flexibility should allow one to account for the effects of ‘idiosyncratic’ ties between origin and destination states (imperfect information). Thirdly, the model attempts to bring together elements from both the neo-classical (economic opportunities) and the cumulative-causation (migrant networks) perspectives on migration. Finally, the model is estimated by using the ‘non-classical’ generalized maximum-entropy method to cope with overparameterization in a cross-sectional setting. An interesting finding is that inter-state migration in Mexico is generally inelastic with respect to geographical distance measure after correcting for the effects of some structural (historical) components of migration, such as migrant stocks.