A Hurdle Model Approach of Store Choice and Market Area Analysis in Grocery Retailing

Abstract

Store choice models play an important role in applied geography, especially in location analysis and retail impact assessment. Econometric store choice models enable the analysis of variables influencing store choice based on observed shopping behavior. Using these models—such as the multiplicative competitive interaction model or the discrete choice model—faces several methodical and conceptual problems limiting their explanatory power and applicability. Addressing these problems, a new approach of store choice analysis is developed, based on a generalized linear model for truncated data, the hurdle model. This approach is applied to grocery stores in Karlsruhe, based on empirical data from a customer survey. Three main insights can be gained: (1) The explanatory performance of the model is superior to the common models as it allows decomposing spatial shopping behavior into a store choice and the intensity of customer–store interactions, such as shopping trips or expenditures. (2) The hurdle approach allows for dealing with a large number of empirical customer–store interactions equal to zero, which is an important statistical feature. (3) The model results confirm previous findings; for example, that spatial shopping behavior in grocery retailing is mainly affected by the accessibility and size of the stores, which proves the approach’s plausibility.