Modelling the attraction of shopping centres

Development of major shopping centres continues even though online shopping is increasing. This has implications for mode and destination choice for shopping travel and therefore also for sustainability. In this paper, we estimate models for shopping travel using an unusually rich data set of shopping attractions. We find that shopping travel is best represented in three separate models: consumables in short and long-activity segments and durables. In all of these models, we show that representing nearby attractions outside the destination zone adds to the measured attraction. For both consumables segments, the addition of secondary attractions within 2 km of the main destination adds very significantly to the explanation given by the model, but attractions within 5 km do not add anything. For durables, both 2 km and 5 km added to the model, but 2 km was slightly better. Furthermore, we find significant within-zone correlation in all three models, but are unable to find significant between-zone correlation, indicating that zone boundaries have some behavioural meaning for shopping travellers, but larger areas are not viewed in this way. Shopping attractions with a specifically Swedish impact, Systembolaget (official alcohol outlet in Sweden) and IKEA, proved to be important in all the models. These attractors work better as part of the size than as part of the utility, indicating that they appear to be separate attractors of trips, rather than as adding to the utility of other attractors.