Spatial interaction models applied to the design of retail trade areas

Intermetropolitan trade areas are geographical zones defined by consumer movements over space -retail flows- from their municipalities of residence towards a head town, to purchase special goods: clothing and footwear, furniture, food, etc. These market areas own an economic sense that do not have other more commonly used territorial divisions, such as towns, provinces or regions. Since 1992, the Lawrence R. Klein Institute -Autonoma University of Madrid-actualises the Spanish Retail Trade Atlas and determines regional trade areas and sub-areas, using spatial gravity models and survey. The authors' experience in this Project allows them to analyse the different procedures suggested for modelling the consumer store-choice process and from this, estimating the market share of a retail outlet or a town. Store choice models can be classified into two main groups. First, the descriptive-determinist approach includes a group of techniques that rely on observation or normative assumptions. It is well-known the procedure devised by Applebaum (1961) for constructing primary trade areas from customers spotted on a location map or the classical central place theory, based on the nearest-centre hypothesis. 'Reilly's law of retail gravitation' (1931) considers not only distance but also attractiveness of alternative shopping opportunities. Secondly, the explicative-stochastic approach uses information revealed by past behaviour to understand the dynamics of retail competition and how consumers choose among alternative shopping opportunities. Huff was the first to use a utility function and introduced the spatial interaction models to explain consumer behaviour. They argued that consumers rate alternatives on the basis of their evaluation of the total utility of the store and not merely on its location. Huff's model is a particular case of the discrete-choice models known as multinomial logit (McFadden, 1974). Both models satisfies the so-called 'Independece of Irrelevant Alternatives' (IIA) property, that is, the ratio of the probabilities of an individual selecting two alternatives is unaffected by the addition of a third alternative. While this may be reasonably representative of certain aspatial choice situations, it is very unlikely to occur in spatial choice because of the fixed locations of spatial alternatives. The competing destinations model, derived from purely spatial considerations, provides a way of overcoming some problems with the logit and nested logit models that arise from the transference of essentially aspatial theory to the spatial realm. This work focuses on market area delimitation models and presents the estimation process developed by the L.R. Klein Institute in determining intermetropolitan trade areas. It is also applied a competing destinations model to the trade area of Madrid, a very peculiar one because of its magnitude and the important shopping concentration around the capital. Finally, we want to highlight the main applications derived from the knowledge and actualisation of the consumer retail flows. These applications take into account not only retailing but also another economic activities relating with market attraction areas.