Purpose: This paper studies how socio-economic factors in the family - namely family income, work hours of the head of household, education of the head of household, and presence of a family member who has ever been incarcerated - affect 17 adolescent risk behavior outcomes, including six delinquency items (hurting someone badly, stealing, damaging school property, skipping school, being stopped by the police, being arrested); six substance use items (ever smoking a cigarette, smoking regularly, drinking alcohol more than once a month, drinking five or more alcoholic drinks in a row more than once a month, ever using marijuana, consuming marijuana three or more times in the past month); and five indices (4-items delinquency scale, 6-items delinquency scale, substance use scale, substance abuse scale, and the behavior problem index). The analysis evaluates whether the four socio-economic variables impact adolescent risk behavior directly, or through familial psychosocial factors (proxied by PCG's psychological distress, self-esteem, and self-efficacy; and a family conflict index), PCG's neighborhood rating, and a peers' risk behavior scale.

Data: The analysis matches adult family members in the Panel Study of Income Dynamics (PSID) to children in wave II of the PSID's Child Development Study (CDS-II), allowing for the study of intergenerational impacts of economic resources, and also for the first time, intergenerational transmission of criminal or anti-social behavior. The sample comprises 1,468 adolescents aged 12 to 19 in 2002.

Methods: The study employs probit regressions on the individual delinquency and substance use items, ordered probit regressions on the delinquency, substance use, and substance abuse scales, and OLS regressions on the behavior problem index. Instrumental variables estimation using census conditions geo-matched to respondents' neighborhoods corrects for endogeneity in the peers and neighborhood variables.

Results: Economic factors have independent effects on youth delinquency and substance use, and also indirect effects through family psychosocial conditions, and peers. Incarceration history in the family proxies for adult criminal behavior, and has weaker direct effects. The attenuation pattern of this variable suggests that it either influences adolescent behavior through other factors or that adult and youth anti-social behavior are predicted by the same factors in the family and neighborhood. Among the family psychosocial factors, the primary care-giver's self-efficacy matters the most to youth risk taking. In direct regressions, peer effects are extremely strong on all measures of risk behavior, but neighborhood effects are weak. For most of the outcomes studied, the magnitudes of the estimates are reduced but still persist after instrumental variables correction, indicating that the initial contrasting results from peers and neighborhood is partially due to endogeneity biases.

Implications: The presence of intergenerational impacts suggests that anti-poverty and anti-crime policies have benefits beyond those that are directly measured because of the spillovers accrued to the next generation. The interaction of economic, family psychosocial, neighborhood, and peer influences imply that effective intervention will need to be targeted at multiple levels.