Choice-based estimation of Alonso’s theory of movement: methods and experiments

Alonso’s theory of movement (ATM) provides a general framework for modeling spatial interaction. Though conceptually elegant, application of this model is limited due to the difficulty in estimating model parameters. In this paper ATM is applied in retail modeling using choice data in a novel way. The approach permits flexibility with regard to model assumptions about flow constraints; that is, the model is neither attraction constrained nor production constrained. Data aggregated to the census ‘block-group’ scale, with network-based travel distances, have been examined to estimate model parameters by pairwise comparisons and ordinary least squares. The expenditure flow is related to the store properties (attractiveness, relative accessibility) and the spatial separation between origins and destinations. Based on our empirical estimate of a key parameter, and assuming that sales increase at a decreasing rate with respect to store size, we deduce that accessibility is inversely related to store size.