The classic model of moral hazard suggests that health insurance may reduce preventive care because the insurer will pay for part of the treatment in case of disease. However, if health insurance covers preventive care as well, the reduced cost of preventive care will encourage the insured to consume more preventive care. These two countervailing effects are referred to as ex ante and ex post moral hazards (Zweifel & Manning 2000). Most studies do not distinguish the two effects, leading to a potentially wrong characterization of moral hazard.

Using Medicare coverage as an example, this thesis identifies ex ante and ex post moral hazard effects of health insurance on cancer prevention. As we know, Medicare eligibility rules increase health insurance coverage at age 65. However, some preventive screenings were not covered in Medicare until recently. The different timing of Medicare eligibility and Medicare expansion of preventive care allows me to use a difference-in-differences framework to separate ex ante and ex post moral hazards.

I focus on female uptake of breast cancer screening and male uptake of prostate cancer screening, using the Medical Expenditure Panel Survey (MEPS) and the National Health Interview Survey (NHIS). In both datasets, I find evidence in support of ex ante and ex post moral hazards. No evidence shows that people try to delay screening until it has been covered by Medicare. Moreover, the level of prevention and responsiveness to insurance changes vary with demographics, with larger effects among whites and the better-educated.

Then I take a second look at the moral hazard problem in the health insurance market using the Health and Retirement Study (HRS). Compared with MEPS or NHIS, the panel nature of HRS allows me to control for individual fixed effects and therefore provides a more stringent test. The major findings on breast cancer screening are consistent. I find strong ex ante and ex post moral hazard effects in it, and individual reactions to Medicare enrollment and Medicare's preventive care coverage vary by factors such as race and income. However, moral hazards on prostate cancer screening is not found, mainly due to data limitation.