Reexamination of the disparity in utilization of medical care services between the insured and uninsured

A lack of insurance coverage has been blamed for the disparity in utilization of medical care services between the insured and uninsured. However, such an argument, may have overstated the effect of uninsurance. Using data from pooled 1996--2000 waves of the Medical Expenditure Panel Survey (MEPS), I reexamine the disparity in utilization of medical care services across insurance status using the following three approaches. Additionally, this dissertation takes into account of the dynamic nature of insurance coverage by distinguishing between temporary (less than a year) and continuous (a year or longer) uninsurance.; First, I hypothesize that the population consists of two (or more) distinct subpopulations, differing in their demand for health services driven by latent health status and risk preferences such that a lack of health insurance coverage may affect their utilization differently. Using a finite mixture approach, I find evidence of discrete heterogeneity in office-based physician visits: only one third of the population is identified as "heavy users" and they are disproportionately (more) affected by a lack of insurance coverage.; Next, I examine the effect of self-selection by jointly modeling insurance decision and health care utilization using a latent-factor-structure model that takes into account selection on both observable and unobservable factors. Results from this study show that part of the observed reduction in utilization among the uninsured can be attributed to selection effect. However, even after controlling for selection, a lack of insurance coverage still significantly reduces office-based physician visits and filled prescriptions, although the effect is much smaller among the temporarily uninsured than the continuously uninsured.; Finally, I test the hypothesis that the temporarily uninsured would engage in utilization timing by shifting doctor consultations from the period of uninsurance to the period of insurance using the unique event history data from the MEPS. Particularly, it tests whether the utilization "speeds up" as individuals expect a termination of the current coverage or after they regain their coverage from an episode of uninsurance. The data provide support for this hypothesis among some "vulnerable" subpopulations.