Are point-in-time measures of neighborhood characteristics useful proxies for children's long-run neighborhood environment?

How well do the point-in-time neighborhood measures commonly used in studies of neighborhood effects represent longer-run neighborhood environment? Our evidence on children's year-to-year correlations in neighborhood characteristics 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. (C) 2002 Elsevier Science B.V. All rights reserved.