A comparison of three models of internal migration in Mexico

Estela Rivero-Fuentes, The Population Council

In this paper I compare the results of applying the three models of migration most commonly used in the empirical literature of internal migration to the case of interstate migration in Mexico. I fit a hierarchical logit model, a nested logit model and a conditional logit model to state-to-state migration flows during 1975-1980, 1985-1990 and 1995-2000. I compare the results of the three models in terms of their conclusions about the factors affecting migration, in terms of their accuracy when predicting out-migration, and in terms of their accuracy for predicting immigration. My results show that the nested logit model is the model that better explains migration during the three periods analyzed. Still, the results of hierarchical logit models do not differ much from the results of nested logit model. Conditional logit models, a method commonly used, yields the worse fit and may lead to erroneous conclusions about the determinants of migration.