Three Essays on Access, Supply, and Policy in Opioid Markets

Abstract

The United States opioid epidemic has claimed nearly one million lives over two decades, yet the forces sustaining it---from illicit supply to treatment barriers---remain incompletely understood. This dissertation examines the opioid market along a connected chain: how much of the opioid supply was diverted to non-medical use, and what happens when enforcement restricts that supply? Does effective medication-assisted treatment reach those who need it? And when a prescription exists, why do patients still fail to fill it? Each chapter addresses one link in this chain, together tracing the epidemic from its supply-side origins through the treatment infrastructure that might contain it. In the first chapter, I study the supply side of the opioid market. Using pharmacy-level opioid shipments across ten U.S. states and a structural model of medical and non-medical demand, I decompose what fraction of dispensed opioids was diverted for non-medical use. I find that 8% of pharmacies dispense opioids for non-medical use with high probability, and that over half of all opioid shipments between 2008 and 2010 were diverted. Critically, aggressive DEA enforcement actions against rogue pharmacies do not eliminate non-medical demand---they displace it. A substantial share of non-medical users migrate to the black market when pharmacy access is curtailed, potentially accelerating the transition to more dangerous substances such as heroin and fentanyl. In the second chapter, I turn to the treatment side of the crisis and ask whether expanding the supply of buprenorphine prescribers saves lives. I exploit the 2016 Comprehensive Addiction and Recovery Act, which for the first time authorized nurse practitioners to prescribe buprenorphine for opioid use disorder, interacting this federal reform with pre-existing variation in state scope-of-practice laws. Granting nurse practitioners independent prescribing authority increased active buprenorphine prescribers by 47%, raised dispensation by 26\%, and reduced opioid-related mortality by 22%. Gains were concentrated in previously underserved counties where no treatment had existed; in counties with established access, dispensation rose but mortality did not improve, with suggestive evidence of diversion into secondary markets. In the third chapter, I investigate why patients who receive a buprenorphine prescription so often fail to fill it. Using administrative insurance data from Washington State, I document that fewer than half of first-time patients successfully initiate treatment---far below fill rates for other chronic medications. I develop and estimate a sequential search model in which patients face both physical travel costs and uncertainty about which pharmacies carry buprenorphine. Search costs account for 70% of treatment failures. Providing prescribers with real-time pharmacy inventory data could raise initiation rates by 17%.