DIRECTIONAL EXPANSION IN MIGRATION MODELLING DIRECTIONAL EXPANSION IN MIGRAnON MODELLING

ABSTRACf Migration is typically considered to be directional, yet our models rarely take into acCOtmt that directionality. Certainly migrants have preference for places in certain directions and show disfavor for those in other directions. The purpose of this paper is to extend the present knowledge of the effect of expansion on migration models. This paper uses an expanded spatial interaction model to elicit the directionality of migration. The results of this model lend insight into the stability of parameter estimates and the regional affiliations of states. Migration research has recognized the directionality in migration for many years. Such concepts as beaten paths, regional shifts, and migration streams all include an implicit directionality. Migration models, on the other hand, have assumed that direction does not matter. The typical gravity (or spatial interaction) model uses distance and population (or attractiveness) as the prime determinates of the flow between two places while the friction of distance and the attractiveness of a place can vary quite significantly over direction. Clark (1986), for example, discussed the regional flows of net migration. He considered the regional shift from the Northeast and Midwest to the South and West However, when he modelled these shifts, the flows return to non-directional movements. The parameter estimates of a model that ignores direction will be unstable if the direction of migration is an important variable. Casetti and Can (1986) have proposed an expansion that will take into accotmt the direction of the migration. It uses a trigonometric expansion of the population and distance parameters. The purpose of this paper is to extend that analysis and len