Does Medicare Reduce Medical Debt?

This paper contributes to the growing literature on the effects of health insurance on individuals’ financial well-being, focusing on the effects of Medicare. As in a number of recent papers that assess the effects of changes in insurance such as the Oregon Medicaid experiment, the Massachusetts health reform and the ACA Medicaid expansion, we make use of administrative data from consumer credit records. Our primary outcome variable is the amount of medical debt sent to third-party collectors, a measure of financial strain. Our work is complementary to Barcellos and Jacobson (2015), who recently studied the effect of Medicare on out-of-pocket medical spending and self-reported proxies for financial strain using survey data. Like them, we employ a regression discontinuity (RD) design, exploiting the program’s eligibility age at 65. Our data come from a nationally representative two-percent sample of U.S. consumers with a credit history, obtained from one of the three major credit bureaus. The data are an unbalanced panel, representing annual snapshots of individual records, culled at the end of August, 2011 through 2016. Our primary sample uses year 2011 through 2013, and includes approximately 1.7 million unique consumers aged 50 to 80. While we do not directly observe the flow of new medical debt in collections, we proxy it from changes in the stock of medical collections over time owed by each consumer in the data, arguing that any discontinuities around age 65 in changes in the stock are likely due to discontinuities in the flow, an assumption we can test to some degree. We hypothesize that if Medicare reduces medical debt, it does so primarily among individuals who are uninsured prior to age 65. While we do not observe insurance status in the credit records, we test this hypothesis using a specification in which the discontinuity is a function of the baseline near-elderly uninsured rate in the consumer's zip code. Because the uninsured rate at age 64 fell substantially after 2013 with the implementation of the ACA, we also test whether the RD around age 65 changed in the later years of our data. Preliminary analysis shows a rather sharp drop in the year-over-year change in medical collections at the mean, beginning primarily at age 66. This timing is consistent with the fact that unpaid bills are not immediately sent to third party collectors, and that there is a lag in reporting of debt in collections to the credit agencies.