We use a natural experiment in Denver to quantify what we argue are causal relationships between rich measures of neighborhood context and asthma diagnoses among low-income minority children. We analyze Latino and African American children from 710 families who resided in Denver public housing during part of their childhood.  Data analyzed come from a retrospective survey of current and former residents of the Denver (CO) Housing Authority (DHA).  Because assignment of households on the DHA waiting list to public housing mimics a random process, this program represents an unusual natural experiment holding great potential for overcoming parental geographic selection bias and thereby permitting drawing causal inferences about neighborhood effects. 

Our goal is to uncover the degree to which the ethnic and nativity mix of a neighborhood, all else equal, have any relationships with a wide variety of outcomes for low-income Latino and African American children and youth.  We stratify our models by Latino/African American groups and thereby investigate the degree to which these demographic compositional elements have similar impacts across these two groups of children.

We employed both logit and hazard models to estimate parameters predicting child and youth outcomes related to physical and behavioral health, exposure to violence, education, marriage and fertility, and employment.  Our focal predictors are the census tract percentages of African American, Latino and foreign-born populations.  We control for a wide variety of other census-based characteristics, as well as administrative data on crime, child abuse and environmental quality and caregiver-reported neighborhood indicators related to social capital, social problems, institutional resources and bad peer influences.  We also control for a range of child, caregiver and household characteristics.  Moreover, because all of our observations come from consolidated Denver City and County, we have implicit controls for political jurisdictional public service package.  

Our results indicate that higher percentages of African American and Latino neighbors are strongly associated both statistically and substantively with a range of adverse child outcomes related to health, exposure to violence, education, marriage and fertility, and young adult education and employment.  On the contrary, higher percentages of foreign-born residents (most of whom are of Mexican origin in our sample) are generally associated with more favorable child outcomes in most domains.  Our results imply that ethnic segregation may impose costs on low-income minority children and youth, above and beyond the associations between segregation and inferior institutional resources and environmental quality.  Policymakers therefore should encourage the development of more diverse neighborhoods according to ethnicity and national origin if they wish to expand opportunities for low-income minority children and youth.