Empirical Tests of an Entropy Maximizing Model of Retail Location and Consumer Behaviour

Abstract

<p>The aim of this research paper is to show that entropy maximising models of urban land use are better than first generation large scale land use models. In order to substantiate this claim it will be argued that entropy maximising models are internally consistent, theoretically based, realistic and parsimonious.</p> <p>First it is shown that an extended family of consistent spatial interaction models can be built using the entropy maximising (information minimising) formalism.</p> <p>The second step is to develop a detailed model. The model presented represents several aspects of urban spatial interaction. In the model, individuals are assigned to place of residence, place of work and shop trip pattern for an arbitrary distribution of service centres. An entropy statistic, defined over the assignment of individuals is used to determine, endogenously, the entropy maximising distribution of centres. Thus, the model predicts the location of retail facilities as well as residential location, shopping patterns and work trip interchanges. The model also has a theoretical basis in that results using a similar formulation for a linear city show that the entropy maximising distribution of service centres includes downtown.</p> Having set up a theoretically based, consistent model, the next step is to establish that this model produces a realistic representation of the city. Data from Hamilton, Ontario are used to test the fit of the model.</p> <p>Results show that the model reproduces the observed data with some accuracy. Furthermore, the endogenously predicted distribution of service centres includes downtown. This result supports the contention that theoretically based entropy models can produce realistic results, and establishes the argument of the paper.</p>