The b2c e-commerce landscape of the Dutch retail sector

Business-to-consumer (b2c) e-commerce can be regarded as a disruptive process innovation that can make existing business models obsolete. B2c e-commerce provides retailers the possibility of a new service concept, new client interface and even delivery system. The history of retailing is replete of such innovations, like the introduction of department stores, mail order etcetera. It is only recently that researchers from various disciplines are examining the way retailers respond to b2c e-commerce as a major new innovation. Despite the growing attention from researchers to the adoption of b2c e-commerce by retailers, there is still little known about rate and extent of this innovation adoption process. Furthermore, studies concerning the diffusion of b2c e-commerce in retailing largely lack a geographical context. In this paper we examine the geographical pattern of b2c e-commerce adoption of shops in the Netherlands. A distinction is made between the two main stages in b2c e-commerce adoption that is the adoption of an active website and online sales. The main hypotheses hold that: (1) population density will positively affect the probability of adoption following the hierarchical diffusion theory and (2) the density of shops in the same sector will positively affect the probability of adoption due to inter-firm competition and imitation. In our analyses, we will control for size, sector and organisational form. For this paper we used a subset of the retail location database of Locatus that consists data of 23,312 shops in the Netherlands, which is 17 percent of all shops in the Netherlands. By a time-consuming procedure we searched for the Internet strategy of the individual shops in our dataset. To improve the accuracy, the data is currently re-examined by two trained coders. The subset contains location data of shops in nine retail categories: supermarkets; drug stores; perfume & cosmetic stores; ladies wear; family wear; menswear; book stores; CD stores; and computers stores that have adopted the Internet. Furthermore, the dataset distinguishes seven types of shopping centres: solitary urban locations; neighbourhood centres; city district centres; city centres; large-scale (peripheral) retail concentrations; business parks; and solitary peripheral locations. Other geographical classifications included in the dataset are: Zip code; municipality; and province. Given the variety of geographical levels in combination with the large number of cases, we will use multi-level analysis to investigate the relevance of geography for retail Internet adoption. We will include three geographical levels in the multi-level analysis: (1) shopping centres, (2) municipalities, and (3) COROP regions or provinces.