Associate Professor of Economics at University of California, Davis
The Long-run Effects From Being Exposed to Troubled Peers
By Scott E. Carrell, UC Davis; Mark H. Hoekstra, Texas A&M; and Elira Kuka, UC Davis
Domestic violence is a significant problem in the United States, with 24.3% of women and 13.8% of men aged 18 and older having been victim of severe physical violence by an intimate partner in their lifetime (Black et al., 2011). These victims incur significant medical and emotional costs (CDC, 2003), and their children exhibit aggressive behavior, depression, decreased social competence and diminished academic performance (Edleson, 1999; Wolfe et al., 2003; Fantuzzo and Mohr, 1999).
Domestic violence not only has negative effects on the families affected by it, but it also has important negative externalities. Carrell and Hoekstra (2010) show that primary school children exposed to domestic violence negatively affect the academic achievement of their peers. Moreover, Carrell and Hoekstra (2012) show that these effects are strong for yet-to-be-reported cases of domestic violence, and that once the violence is reported these negative spillovers disappear.
While the evidence of the negative spillovers from domestic violence on academic achievment is strong, little is known whether these effects are temporary or persist in the long run. Our results show that exposure to a higher proportion of peers in primary school from families affected by domestic violence leads to lower academic achievement in secondary school. Moreover, we find that these negative effects are strong and significant for yet-to-be-reported cases of domestic violence, and statistically close to zero for cases that were previously reported by the parent
We also analyze the effects of troubled peer exposure on secondary school completion, college enrollment and degree attainment. We find that a 1 percentage point increase in peers with yet-to-be-reported domestic violence leads to a 0.5 percentage point decrease in the probability of enrolling in college and in the probability of obtaining a Vocational or Associate degree. As with academic achievement, we find these effects to be small and statistically insignificant for already reported domestic violence.
To overcome self-selection into schools and classrooms, we exploit the variation in peer domestic violence that occurs at the school-grade-year (cohort) level, while controlling for school-grade specific fixed effects. Thus, our identification strategy relies on idiosyncratic shocks in the proportion of peers from families linked to domestic violence, across grade cohorts, within a school, over time.