Three essays on health insurance and health care consumption

The first essay of this dissertation aims to study the decision to enroll in supplementary insurance plans for the Medicare elderly. It also examines the impact of insurance status on several health care utilization measures. The insurance decision and utilization can be influenced by common unobserved attributes, such as attitude toward risks. Therefore, I model medical utilization and insurance choices jointly with a latent factor specification in order to control for unobserved heterogeneity. The estimation method is maximum simulated likelihood. This chapter analyzes a five-wave panel data set derived from the Health and Retirement Study. The empirical results indicate a significant positive relation between supplementary insurance and nights spent in a nursing home, nursing home admission frequency, and hospital admission frequency. The evidence of selection on unobservables is modest. The results also show that supplementary insurance has a negative impact on number of doctor visits in clinics, a measure which excludes doctor contacts made in a hospital or nursing home. In addition, the number of doctor visits exhibits high positive correlation with unobserved heterogeneity.; The second essay studies the impact of various supplementary insurance plans on medical expenditures of the Medicare elderly, with a focus on total out-of-pocket expenditure on all kinds of medical services used and out-of-pocket expenditure on prescription drugs. I use quantile regression method and finite mixture models to account for heterogeneity in medical consumption. The empirical results demonstrate the distinct impact of different supplementary insurance plans on health care expenditures. Employer-sponsored supplementary plans have a significant negative impact on total out-of-pocket, and out-of-pocket drug expenditure, relative to individually purchased plans; HMO plans decrease out-of-pocket expenditure significantly relative to other private plans. Quantile regressions show mild variation in the impact of supplementary insurance choices across quantiles. The results from finite mixture models support the hypothesis which assumes the existence of two latent subpopulations. That is, the impact of insurance choice is quite different in the heavy users group and the light users group.; The third essay investigates the switching behavior of non-elderly enrollees in U.S. managed care plans. Treatment effect analysis is used to examine the disaggregated expenditures of plan switchers and plan stayers prior to their decision to switch or stay. Propensity score matching methods are used to estimate the average treatment effects on the treated. The results, which are based on a national representative data set from the Medical Expenditure Panel Survey, indicate that switchers (from HMO to non-HMO) spend more on hospitalization. The other type of switchers (from non-HMO to HMO) spends less on prescribed medicine and office-based physician visits. The findings suggest that the non-HMO private managed care plans provide better coverage on hospitalization, office-based physician visits and prescribed medicine than the HMO plans.