Are Single-Year Measures of Neighborhood Characteristics Useful Proxies for Children’s Long-Run Neighborhood Environment?

Numerous studies have estimated the degree to which children’s neighborhood environment influences their subsequent socioeconomic status. Most of these studies identify the child’s neighborhood at a single point in time. This could lead to underestimation of neighborhood effects if point-in-time neighborhood measures are poor proxies for a child’s longer-run neighborhood environment. In this paper, we investigate the appropriateness of using neighborhood characteristics at a point in time by looking at children’s year-to-year correlations in neighborhood characteristics. High autocorrelations would suggest that point-intime measures are likely to be good proxies for a child’s longer-run environment. Low autocorrelations would indicate that estimated neighborhood effects based on the neighborhood inhabited at one point in time may be subject to substantial downward bias. We find that children’s neighborhood characteristics are highly correlated across years. This suggests that the neighborhood a child inhabits at a particular time is a reasonable proxy for her long-run environment, and that relying on such a proxy produces only a small errors-in-variables bias.