Three essays on family structure and school dropout among adolescents

This dissertation is a collection of three essays on high school dropout by adolescents and the roll of their siblings in this risky behavior. Dropping out of high school can scar the individual for his entire labor supply period with lower earnings, higher unemployment, and in turn incur considerable social and welfare costs to the economy. The data used in this study is from National Longitudinal Survey of Youth 1997, which has information on about 8,900 individuals aged 12-17 in the year 1997 who are followed up every year since. I use the first 10 rounds of this survey. The first essay in this dissertation identifies the influence of older siblings on their younger siblings’ decision to leave school before graduation. It starts by representing the significant adverse effect of birth order on school completion outcome of the teen. The results indicate that teens with older siblings are more at risk of dropping out of school. Having these initial estimates of the siblings’ effect, I look at the behavior of the siblings pairs in the in a subset of 519 pairs from families with two children for which the data is available on both siblings’ outcome and characteristics. I address the question of the inherent endogeneity by using family fixed effects models on these pairs, and also by implementing instrumental variables technique. I exploit the older sibling’s specific characteristics as an instrumental variable for his outcome in the younger sibling's equation. The sibling specific characteristic used as instruments are: the older sibling’s gender, the unemployment rate when the older sibling was 16 years old, the mother’s age at his birth, and the intactness of his family when he was 14 years old. The results of the estimation using this set of instruments show positive and significant effect on early school leaving of the younger sisters. The second essay utilizes a survival model to determine the timing pattern of teenager’s decision to drop out. Preventing and intervening in school dropping out in teens require the knowledge of the timing and pattern of occurrence of this act. I use nonparametric, semiparametric and parametric hazard models to explain the factors affecting the age onset of high school dropping out. Alongside with other socioeconomic factors, this chapter reinforces the roll of siblings on reducing the starting age of this risky behavior in the younger siblings. The results show that a teenager who has an older sibling is more likely to stop schooling at a younger age. In teenage boys who have an older sibling the hazard of dropping out is about 3 times as much depending on the choice of the hazard model for the time to first dropout. The shared frailty among siblings of the same family that contaminates the estimates is measured for the parametric and semiparametric duration models using the expected-maximization algorithm for the clustered data. The impact of shared frailty is ruled out in the case of Cox semiparametric models, but had been evident in some of the parametric models used in this chapter. The third essay develops a binary choice interaction model with finite number of agents to characterize the peer effect of siblings on the strategic choices of the teenager. This dynamic model incorporates the attractiveness of imitating the behavior of the peers inside the family. The model measures the strategic complementarity between the choice of the teenager and the choice of his or her siblings, after controlling for shared family fixed effects and utilizing the lagged dependent variable to reduce the unobserved contextual and correlated effect. This model finds significant social interaction effects between siblings, and specifically siblings that are closer in age to one another. The broad policy implication of this research is to indicate another important channel through which the strategic planning to reduce school dropout rate could be directed. School dropout prevention programs can put more emphasis on the first order or lower order children in multi-kid families to utilize the existing spillover effect on the younger siblings. Also considering this spillover effect, parents’ investment in raising a more scholarly firstborn might help them get an additional indirect return to their investment benefiting the other kids in the family.