This dissertation consists of three essays on health economics. The first chapter provides empirical evidence on the impacts of government reimbursement of long-term care. We apply a regression discontinuity design using administrative data from South Korea to estimate the impact of subsidies for formal home and institutional care on informal care use and medical expenditures. We find that reimbursement leads to increases in formal long-term care utilization, even accounting for crowd out of private spending. Among individuals who are partially dependent for some activities of daily living (ADLs), we find that increased use of formal home care has no impact on the use of informal care at the extensive margin or on medical expenses. Among individuals who are partially dependent for several ADLs, we find that increased use of institutional care leads to reductions in informal care and medical expenses. Among individuals who are completely dependent for several ADLs, we find that substitution of home care for institutional care leads to substantial decreases in medical spending. The second chapter studies state laws passed in the late 1990s that required health insurers to cover diabetes related equipment, supplies, and education. We assess the impact of these mandates on health related behavior and labor market outcomes. We find no significant effects for diabetics or groups with higher prevalence of diabetes in terms of exercise, diet, income, or employment. These results are robust to different specifications and datasets. The third chapter provides empirical evidence on both outcomes and potential mechanisms resulting from information obtained from screening. We apply a regression discontinuity design using administrative data from South Korea to estimate the impact of different classifications of overall health that vary discontinuously with blood sugar level. We find that secondary examinations due to a "disease suspected" classification leads to follow-up rates greater than 50%. However, we find few impacts otherwise, including short and medium run medical activity and longer run health outcomes. We also find that the responsiveness to the classifications among the highest income quintiles is lower than among the other quintiles, consistent with more educated individuals incorporating information directly from the blood sugar measure itself.