UC Irvine

"Prior research frequently observes a positive cross-sectional relationship between various neighborhood structural characteristics and crime rates, and attributes the causal explanation entirely to these structural characteristics.  We question this assumption theoretically, proposing a household-level model showing that neighborhood crime might also change these structural characteristics.  We test these hypotheses using data on the census tracts in 13 cities over a ten-year period, and our cross-lagged models generally find that, if anything, crime is the stronger causal force in these possible relationships.  Neighborhoods with more crime tend to experience increasing levels of residential instability, more concentrated disadvantage, a diminishing retail environment, and more African Americans ten years later.  Although we find that neighborhoods with more concentrated disadvantage experience increases in violent and property crime, there is no evidence that residential instability increases crime rates ten years later."