Exploring the spatial segmentation of housing markets from online listings

TL;DR

This paper presents a network-based methodology to identify spatial segmentation in housing markets from online listings, revealing robust, country-specific submarkets relevant for urban planning and policy.

Contribution

Introduces a multipartite network approach to detect spatial market segmentation from online housing data, addressing a key challenge in urban studies and spatial economics.

Findings

Market segmentation is consistent across different clustering methods.

Spatial units form coherent, country-specific submarkets.

Method is robust across spatial scales and data discretizations.

Abstract

The real estate market shows an inherent connection to space. Real estate agencies unevenly operate and specialize across space, price and type of properties, thereby segmenting the market into submarkets. We introduce here a methodology based on multipartite networks to detect the spatial segmentation emerging from data on housing online listings. Considering the spatial information of the listings, we build a bipartite network that connects agencies and spatial units. This bipartite network is projected into a network of spatial units, whose connections account for similarities in the agency ecosystem. We then apply clustering methods to this network to segment markets into spatially-coherent regions, which are found to be robust across different clustering detection algorithms, discretization of space and spatial scales, and across countries with case studies in France and Spain. This methodology addresses the long-standing issue of housing market segmentation, relevant in disciplines such as urban studies and spatial economics, and with implications for policymaking.