Efficiency Implications of Households’ Choice of Stores in Grocery Retailing

Modelling Shopping Frequencies

Abstract

The demand for transportation services is generally considered to be induced from an underlying activity, such as commuting to work, performing a leisure activity, visiting friends or relatives, or shopping. According to this line of argument, transportation from origins to destinations is complementary with respect to the activities that are linked together and not isolated phenomena. Shopping frequencies are an important part of a comprehensive economic welfare analysis of grocery store location, due their consequences for the traffic volumes. If households want to purchase groceries with high frequencies, a dense net of stores is efficient. Lower frequencies imply fewer and, consequently, larger stores, where economies of scale within firms are used extensively. Shopping frequency is an example of count data and can thus be expected to follow a Poisson distribution. Therefore, a Poisson regression model is estimated, which expresses how a set of explanatory variables influences frequencies. The set of explanatory variables consists, for example, of socio-economic variables, and type of housing and travelling behaviour. In the material, data about frequencies are available only as grouped data, which suggest that limited dependent models with censored data are a suitable approach. Parameters are obtained through maximum likelihood estimation. The results are promising and can be used as input in a comprehensive appraisal of distribution system efficiency. Lowering the frequency will reduce the total number of shopping trips, hence releasing capacity for other usage in the urban transport system, and enhance a distribution system with large stores. On the other hand, fewer shopping trips necessitate more purchase planning and storing of groceries by households. Moreover, the results will be of interest when estimating the potential of shopping on the Internet, which is also discussed in the paper.