Publication Date



Open access

Embargo Period


Degree Type


Degree Name

Doctor of Philosophy (PHD)


Economics (Business)

Date of Defense


First Committee Member

Philip K. Robins

Second Committee Member

Carlos Flores

Third Committee Member

Laura Giuliano

Fourth Committee Member

Michael T. French


The number of uninsured children in the United States has declined since the late 1990s. According to the National Health Interview Survey, the percentage of children who were uninsured at the time of the survey decreased from 13.9% in 1997 to 7.0% in 2011. Nevertheless, 5.3 million children lacked health insurance coverage in 2011. Among uninsured children, adolescents tend to be uninsured at higher rates than younger children. The effects of health insurance coverage on children have been widely studied in economics and health care policy literature. Findings have indicated positive correlations between health insurance coverage and children’s health care access, health outcomes, educational performance, and quality of life. Few studies have focused on the effects of health insurance in adolescence, a transitional period marked by rapid physical and intellectual growth. For adolescents, lack of health insurance could decrease current well-being and cause negative effects in adulthood. Therefore, it is crucial to study the effects of health insurance coverage for adolescents, which as the results of this study show, can provide insight into the implications for future well-being as adults. In this study, I estimate the effects of health insurance coverage for adolescents on future well-being—including health status, educational attainment, and labor market performance. These three achievement outcomes are treated as response variables in the econometric models built for this study, while health insurance coverage of adolescents is the main explanatory variable in the models. To estimate the econometric models for measuring the effects of health insurance coverage for adolescents, I use a longitudinal survey data set (Add Health), which includes a cohort of adolescents from the 1994–1995 school year who were again studied in young adulthood through follow-up surveys in 2001–2002 and 2007–2008. One of the benefits of using the Add Health survey data in the analysis is that they combine longitudinal data on adolescents’ health insurance, social economic status, parental characteristics, and future well-being. Therefore, the Add Health survey data provide plentiful information on the control variables in the models and help to identify the causal effects of health insurance coverage for adolescents after controlling for these variables. In estimating the econometric models, I first apply ordinary least square (OLS) regression. I also use ordered logistic regression and then compare the results with those of the OLS. Consistent results between the two estimations indicates that after controlling for parental, adolescent, and young adult characteristics, adolescents who are covered by health insurance have significantly higher educational attainment and personal earnings in young adulthood than adolescents who are not covered. To test the consistency of least squares estimates in case there is any endogeneity in the developed econometric models, I use the Durbin–Wu–Hausman test (augmented regression test). Based on the results of the test, I reject the hypothesis that the potential endogeneity problem could be ignored in the models. In addition, I use bivariate probit analysis to test for endogeneity. The results imply that the residual was correlated with health insurance coverage of adolescents and health outcome, which could lead to selection bias. To control for bias in the models, I use the two-stage least squares (TSLS) method. In the first stage of the TSLS method, I construct an instrumental variable and use it to predict health insurance coverage. In the second stage, I estimate the models with achievement outcomes as dependent variables and predict health insurance coverage and control variables as independent variables. The main results of the TSLS analysis suggest positive correlations between health insurance and future outcomes for adolescents, which have sign consistent with the OLS and ordered logistic estimates. Having health insurance could lead to the increase of future education attainment by 14.4% (OLS) and by 140.9% (TSLS). Adolescents who have health insurance tend to have $1648 more (OLS) or $17044 more (TSLS) personal earnings than those who do not have. I also conduct propensity score matching (PSM). Overall, the empirical analysis results suggest the importance of having health insurance for adolescents, which enables them to improve their education and socioeconomic status in young adulthood.


Health Insurance; Adolescents; Future well-being; Endogeneity; Instrumental variable