Abstract

Urban sprawl is a phenomenon observed in most cities around the globe and especially in Latin America, where it is associated to socioeconomic segregation. In the case of Chile, sprawl has been generally based on large real estate projects. Developers target their projects to different types of consumers, which translates into submarkets with a broad range of housing-unit's characteristics, but also different location strategies. This heterogeneity has been analyzed and measured in the literature, but quantitative studies have used exogenous or sequential methods to identify submarkets, leading to potential bias in the segmentation. In this paper, we propose an econometric model to measure location drivers for different types of real estate projects that fills this gap. The modeling framework is based on discrete-choice and latent-class models, allowing us to simultaneously identify market segmentations, and their particular location choice preferences, without the need of arbitrary or ex-ante definitions of submarkets. The model is applied to the city of Santiago, Chile. The results reveal two clearly different approaches taken by developers to produce housing, with one submarket of "exclusive" and more sprawling projects, and another submarket of "massive" and more density driven projects. Location strategies are very different between submarkets, reproducing the socio-spatial segregation already observed in the consolidated city.