A small body of research finds that the children of owners, compared to renters, are less likely to experience negative outcomes such as school dropout or teen parenthood.  This line of research, however, has not yet identified what it is about homeownership that seems to be good for children.  Since the costs of moving are much lower for renters than owners, renters move more frequently.  Stability, then, is likely particularly important for child outcomes regardless of tenure, but may also account for differences in child well-being between owners and renters.  Using longitudinal data from the National Longitudinal Survey of Youth (NLSY) and Hierarchical Linear Models (HLM), we estimate the effects of  stable housing tenure (owning or renting) and exits from either arrangement on changes in children’s achievement (math and reading tests) and problem behavior (internalizing and externalizing behavior problems). We consider whether the effects of stable tenure and exits vary by children’s ages and assess whether these effects are transitory or persist over time.  We expect that the age of the child is particularly consequential given the marked changes occurring in child development and parental oversight between the ages of 5 and 17.  Younger children spend more supervised time in the home, suggesting that the stability of that arrangement rather than the tenure status might be most important, however, older children with more extensive peer networks and active use of public goods may be impacted by tenure if that status is associated with higher quality schools and other public goods.

An important methodological concern is that families who are less stable regardless of tenure may be different from those who are stable owners or renters, and these differences may, in part, explain their children’s lower achievement and higher levels of problem behavior. Consequently, it is important to use analytic approaches which reduce the likelihood of selection bias. Because the HLM models identify the effects of tenure and exits from either housing arrangement on changes in achievement and behavior, this method reduces bias from unobserved persistent child and family characteristics.  We will also use an instrumental variables strategy to estimate the effects of homeownership exit on child outcomes.  We will use zip-code level foreclosure rates to instrument (predict) overall exits from homeownership. We will then use the instrumented (predicted) probabilities of these events to produce unbiased estimates of the local average treatment effects of homeownership exits.  By considering the impact of tenure and exits on child outcomes over time, we will have direct evidence about what appear the most important drivers and, therefore, what amendments to housing policy might be most effective for children.