How Does Medicare Eligibility Affect Choice and Utilization?: Evidence from Veterans

Access to public health insurance programs has been shown to affect patients’ health care utilization (e.g. Finkelstein et al. (2012)). However, less is known about the effects of increased access (choice) for those who already have generous insurance coverage. We address this question by analyzing the health care utilization patterns of Veterans who have health insurance through the Veteran Health Administration (VHA) and who become eligible for Medicare at age 65. Specifically, we use a regression discontinuity approach and VHA claims data to estimate the change in health care utilization, across many medical procedures, for those who are close to the Medicare eligibility threshold of 65 years. By focusing on this group, we learn about how the onset of Medicare eligibility and the associated benefits, which include increased provider access and choice, matter for utilization. Additionally, by focusing on those veterans who are near age 65, we are able to capture the causal effect of Medicare eligibility on utilization. In fact, our preliminary estimates show a sharp decline in VHA inpatient and outpatient encounters once Veterans reach the age of 65.

This setting is particularly interesting because VHA enrollees do not lose their previous medical benefits upon joining Medicare; rather, the set of providers from whom they can choose is expanded. This is in contrast to prior studies that consider the effect of insurance eligibility on outcomes for those who either previously did not have insurance coverage or for those who simultaneously experienced a change in their provider choice set. Our setting is also interesting because the additional health care options that Veterans gain once Medicare eligible are typically closer to Veterans’ homes, but at a higher out-of-pocket cost to the Veteran. Additionally, the analysis can also help us learn about how patients substitute medical services across public programs. This can be used to better understand the amount of “crowd-out” that occurs across public health insurance programs, which is relatively unknown.