Unlocking neighborhood density

Abstract

Studying the components of neighborhood population density reveals a complex picture that little is known about. Hidden under the same level of population density, neighborhoods can vastly differ in crowding, if residential coverage or building heights are moving in opposite directions. We study this heterogeneity in density components and how it is linked to the variation in neighborhood socio-economic characteristics that define modern cities. To do so, we use novel high-resolution (10 m $\times$ 10 m) geo-spatial data on building height and footprints in combination with Norwegian register data. This data allows us to decompose the variation of density into its components, as well as along various margins. We identify urban spatial structures with a latent profile analysis. These data-driven density profiles turn out to be strongly associated with the sorting of people by socio-economic characteristics, such as income and demographic variables. Our results show that below the surface of density, there is the so-far unknown potential to learn about the prevalence and geography of socio-economic groups in the absence of micro-level data.