Firm relocation patterns incorporating spatial interactions

This paper developed a relocation pattern for analyzing decisions regarding the relocation process of individual firms by using discrete choice models. Within this framework, three decision points in the relocation process are maintained in the micro-simulation modeling. The first decision that is move or non-move is modeled by using a binary logit form with the outcome of the probability of moving. The second decision, choosing the destination region, is modeled as a multinomial logit model with the outcome of the conditional probability of choosing a region. The final decision, choosing the new zone, is estimated by considering the spatial interactions. For example, the correlation among firms in the deterministic term and the spatial correlation among zones in the error term are considered by mixed logit models. The results of the study indicated that large firms and older firms have a lower probability of relocation. Moreover, this research found out that migrating firms are more attracted in regions with high accessibility. Finally, spatial factors and land prices in a given zone strongly affect the zone choice decision-making process of all firms.