Fast and slow change in neighbourhoods: characterization and consequences in Southern California

Due to data limitations, most studies of neighbourhood change within regions assume that change over the years of a decade is relatively constant from year-to-year. We use data on home loan information to construct annual measures of key socio-demographic measures in neighbourhoods (census tracts) in the Southern California region from 2000 to 2010 to test this assumption. We use latent trajectory modelling to describe the extent to which neighbourhood change exhibits temporal nonlinearity, rather than a constant rate of change from year to year. There were four key findings: (1) we detected nonlinear temporal change across all socio-demographic dimensions, as a quadratic function better fit the data than a linear one in the latent trajectories; (2) neighbourhoods experiencing more nonlinear temporality also experienced larger overall changes in percent Asian, percent black, and residential stability during the decade; neighbourhoods experiencing an increase in Latinos or a decrease in whites experienced more temporal nonlinearity in this change; (3) the strongest predictor of racial/ethnic temporal nonlinearity was a larger presence of the group at the beginning of the decade; however, the racial and SES composition of the surrounding area, as well as how this was changing in the prior decade, also affected the degree of temporal nonlinearity for these measures in the current decade; (4) this temporal nonlinearity has consequences for neighbourhoods: greater temporal nonlinear change in percent black or Latino was associated with larger increases in violent and property crime during the decade, and the temporal pattern of residential turnover or changing average income impacted changes in crime. The usual assumption of constant year-to-year change when interpolating neighbourhood measures over intervening years may not be appropriate.