Measuring Cities' Neighborhood Demographic Change as the Movement of Emergent Boundaries

Jonathan Tannen, Princeton University

Cities' spatial patterns in ethnicity and race can be characterized by large clusters of blocks with similar composition and sharp boundaries between them. While most research uses fixed boundaries, such as Census tracts, I argue that boundaries are not predetermined, but emerge endogenously and can move over time. If we allow for moving boundaries, a city's spatial demographics can change in two distinct ways: within-cluster changes in demographic composition and boundary movements. I develop a Bayesian algorithm to identify spatial clusters of households' race and ethnicity through time from block-level Census data, and examine changes in those ethnoracial clusters in Philadelphia, PA from 2000 to 2010. I find that boundary movement played a substantial role in Philadelphia, including being responsible for all of the change in pockets where the White population increased. In these areas, predominantly-White clusters are ``spreading'', rather than the ethnoracial composition within fixed boundaries changing.