Estimating sibling spillovers in health: Evidence on symptoms

This paper estimates the sibling spillover effect in health symptoms using a sample of US adolescents from the National Longitudinal Study of Adolescent Health dataset. The research design of this paper is to restrict the sample to sibling pairs who are separated between schools, where one enters high school and the other middle school. Because of school separation, sibling pairs face independent health shocks from own school peers. The identification strategy further exploits variations in individual health across symptoms to control for unobserved individual heterogeneity, which flexibly partials out family correlated effects. Estimation results show that the sibling spillover effect is large as a one-standard-deviation increase in one sibling's frequency of developing a stomach ache or a loss of appetite increases the other sibling's frequency of having the same symptom by about 55% of a standard deviation. Further investigation suggests that the effect is not due to spillovers in drinking alcohol or depression, but probably due to the spread of contagious illnesses like the stomach flu.