Establishment location choice model considering intra-firm interactions

Abstract

This paper describes the development of a business establishment location choice model for the Greater Toronto and Hamilton Area (GTHA) in Canada. The location of establishments is the origin or destination of freight trips and passenger trips (e.g., shopping trips, employment location etc.). The conventional treatment of establishments is that they act as independent decision-making agents when determining where to locate. However, it is known that many establishments are members of larger firms making strategic location decisions among multiple regions and locations within said regions. The location choice decision is often made to reduce logistics costs, travel time, and maximizing market coverage. Hence, depending on the type of the firm, there could be a significant proportion of intra-firm freight trips, which should be distinguished from inter-firm freight trips. In this study, we focus on location choice of establishments through the lens of firm membership. Firm membership is identified for establishments in the retail and wholesale trade sectors and used to calculate intra-establishment distances. The model is expanded to include intra-firm panel effects to account for correlations in the decision-making of establishments. It is found that establishments in the wholesale industry tend to locate in lower population density areas partially due to their larger land footprints while retail establishments that sell everyday goods such as grocery stores tend to locate in high population density areas. A key finding is the distinction between intra-firm and inter-firm distance preferences. It is demonstrated that establishments from the same firm tend to strategically locate farther from one another, while still choosing to co-locate with other establishments from the same industry. This result highlights a gap in existing methods of analysis, an inability of these models to distinguish between spatial considerations on establishment location choice. We propose the development of choice models capable of representing such dynamics.