Research

Working Papers

“The Health Costs of Cost-Sharing” with Amitabh Chandra and Ziad Obermeyer, NBER WP #28439, Accepted, Qurterly Jounral of Economics

Abstract: The design of health insurance can cause large, abrupt, and arbitrary changes in drug prices. We use this fact to study what happens to patients’ health when they stop taking their drugs. Thanks to a quirk in Medicare’s drug coverage, 65-year-olds are as-if-randomly assigned a budget—a spending limit on drugs, beyond which they are responsible for 100% of the cost—as a function of their birth month. We find that an exogenous $100 per month decrease in this budget (a 24.4% change) causes mortality to increase by 0.0164 percentage points per month (13.9%). This estimate is robust to a range of falsification checks, and in the 97.4th percentile of 541 ‘placebo effects’ formed in settings that are observably similar, but lack the policy quirk linking birth month to drug budgets. We make sense of this large effect in three ways. First, patients stop taking drugs that not only appear ‘high-value’ (e.g., blood pressure medications), but are also known to have withdrawal or ‘rebound’ effects. Harm from abruptly stopping these drugs can be large, dwarfing any foregone benefits of not taking the drug. Second, using machine learning, we identify patients at the highest risk of drug-preventable adverse events (e.g., heart attack). Contrary to the predictions of some economic models of behavior, high-risk patients cut back more than low-risk patients on precisely those drugs that would benefit them the most (e.g., statins). Third, patients appear largely unaware of the risks. In a survey, we find only one-third believe that missing their drugs for up to a month could have serious health consequences. We conclude that, far from curbing waste and moral hazard, cost-sharing causes patients to miss opportunities to purchase health at low cost ($11,321 per life-year).

Press: Vox, The Weeds


Research in Progress

Heterogenous Effects of Mergers in Consumer Packaged Goods

Growth Through Entry Deterrrence: Evidence From Craft Beer Acquisitions