Two essays on effect of general health checks and self-selection bias

Abstract

General health checks have long been a fundamental part of preventive medicine in many countries. This dissertation consists of two essays focusing on measuring the effect of general health checks on various health outcomes, using a large 12-year panel dataset provided by the Korean National Health Insurance Service. First essay proposes an empirical approach which jointly models the behavior to obtain general health check and healthcare utilization, with accounting the self-selection problem by using the eligibility to obtain the general health check for free as an instrumental variable. Correlations between the behavior in taking the first health check and the healthcare utilizations are modeled to address the self-selection issue in observational studies. We find that participation in the general health check increases healthcare utilization and ignored self-selection generates substantial upward bias in the estimates. We also find that the health check effect shows remarkable heterogeneity across gender and income groups. Second essay examines the effect of undergoing general health checks on the treatments of two chronic diseases, diabetes and hypertension. To account for the self-selection problem of participating general health checks, we apply a bivariate probit model with an instrumental variable, leveraging the exogenous variations in the eligibility for free general health checks. We find that a significant self-selection bias exists and the effects of general health checks on detecting two chronic diseases are substantially overestimated if ignored. We also find that self-selection biases are more severe for people with high income and that general health checks are effective in detecting the two chronic diseases among people with low income.