Spatial characteristics of housing abandonment

•Abandoned homes and the predictors of abandonment cluster.•Surrounding neighborhood conditions impact the odds of a house being abandoned.•Some factors around neighborhoods have stronger effects than internal conditions.•LISA maps may provide policy makers with insight on where to spend resources.

This study investigates whether the probability of housing abandonment is influenced by spatial factors. Using indicators of spatial autocorrelation, the study finds that housing abandonment and the predictors of abandonment exhibit statistically significant clustering. More importantly, by comparing a multilevel regression model that does not account for spatial relationships to one that does, the study finds that accounting for spatial relationships significantly improves the ability to predict abandoned housing. Additionally, the study shows that in some cases, conditions in surrounding neighborhoods have a greater influence on the probability of housing abandonment than do conditions in the neighborhood itself.