Working Papers
When Uber Comes to Town: Spatial Frictions and Healthcare Access (Job Market Paper)
American Society of Health Economists Program Chair Award 2025
Abstract: Health care is inherently local, and geographic distance and transportation constraints create spatial frictions that limit access to care. This paper examines how spatial frictions, which endogenously arise from the geographic sorting of both patients and providers, shape health care access and affect patient welfare. I first exploit Uber’s staggered market entry in the U.S. as an exogenous shock to patient travel cost in a difference-in-differences design. Using the universe of administrative claims data on Medicaid-enrolled children, I document two mechanisms through which travel costs shape health care access: how much care is consumed (market expansion effect) and which provider is chosen (substitution effect). Lower travel costs increase patients’ use of primary care visits by 4.6%, which in turn improves key health inputs, such as vaccine adherence. It also changes where care is obtained, influencing provider quality: patients travel 5.7% farther in driving time to new providers, increase visits to doctors in areas with poor public transit, and shift toward pediatric specialists over general practitioners. To quantify the welfare impact of spatial frictions, I estimate a structural model of primary care provider choice with latent choice constraints that account for providers' heterogeneous willingness to accept Medicaid patients. I simulate counterfactual policies that reduce spatial frictions through two channels: 1) transportation subsidies that lower the cost per mile of travel and 2) increased provider payment that expands the network of willing providers, thereby reducing the distance patients must travel. The welfare gains from these policies are heterogeneous across neighborhoods: patients in low-income areas need a combination of transportation subsidies and payment increases, whereas increasing provider payment alone is more effective in middle-income areas. These results suggest a place-based policy design to expand Medicaid access efficiently.
Managing Drug Benefit Through Administrative Burden: Evidence from Medicaid (with Abby Alpert) [Draft Coming Soon]
Abstract: Means-tested programs that allocate benefits to low-income individuals rely on administrative costs (ordeals) as rationing devices. We study the impact of one post-enrollment ordeal in Medicaid: prior authorization. Prior authorization could lower spending through two channels: limiting the set of patients eligible for treatment (screening effect) and increasing the hassles of prescribing by requiring a prior authorization request (hassle cost effect). We study the effects of prior authorization requirements on patient access to prescription drugs, focusing on ADHD treatments in the Pennsylvania Medicaid program. We introduce novel plan-level data on prior authorization criteria and use a Regression-Discontinuity (RD) research design leveraging age discontinuities in prior authorization restrictions. We find that prior authorization on ADHD drugs reduces spending by 13.7%, reduces the number of prescriptions filled by 14.7% and deters drug fills. We also find that both screening and hassle cost effects contribute to the reduction in use, with hassle costs accounting for a larger share. There are limited spillovers in prescribing practices across patients facing different prior authorization restrictions.
Taxing Prescription Drugs: Evidence from New York Opioid Tax (with Jackson Reimer) [Preliminary draft available upon request]
Abstract: Over-consumption of compulsive goods is often addressed by policymakers through corrective excise taxes. While sin taxes for prescription opioids are less common, they have been proposed at the state and federal levels in an effort to curb drug overdoses. The impact of such a tax on opioid use depends on how insurers pass through the cost to patients' out-of-pocket prices. Its welfare implications are also ambiguous because any reduced utilization may reflect declines in both clinically appropriate and inappropriate use. We study New York State's prescription opioid excise tax implemented in 2019, the first of its kind in the United States. The tax varies according to a drug product's morphine milligram equivalent, such that highly potent opioids are taxed more heavily than less potent opioids. Leveraging this variation in tax exposure, we find that the policy reduced opioid utilization among Medicaid beneficiaries. This reduction occurs through substitution from higher-dose opioids to lower-dose, less tax-exposed alternatives. We also find the policy slightly reduced prime-age county employment-to-population ratios. We do not observe an impact on drug-related deaths.
Trends in Reporting and Representation in US Alzheimer’s Clinical Trials, 1997-2023. (with Zhuoer Lin, Joseph Ross, Kien Lau, Sophia Stumpf and Xi Chen) [Link] [Submitted]
Selected Work-in-Progress
Do Regulated Price Cuts Expand Access? (with Hanming Fang, Gordon Liu, Huyang Zhang, Karen Zhang)
The Short-run and Long-run Impact of Regulated Price Cuts on Competition. (with Hanming Fang, Ming Li)
Publications
Occupational Differences in the Effects of Retirement on Hospitalizations for Mental Illness: Evidence from Administrative Data (with Tianyu Wang, Xi Chen and Jody Sindelar.) Economics & Human Biology, 53, 101367, 2023. [Link] [IZA No. 16545] [Media coverage: Yale Institute for Global Health, Yale Center Beijing (in Chinese)]
Small Money, Big Change: The Distributional Impact of Differentiated Doctor Visit Fee on Primary Care Utilization (with Tianyu Wang, Ke Wen and Qiuming Gao) Social Science & Medicine, 339 (2023) 116355, 2023. [Link]
Does Extreme Temperature Exposure Take a Toll on Mental Health? –Evidence from China Health and Retirement Longitudinal Study (with Yanran Chen, Xuezheng Qin and Xi Chen), Environment and Development Economics, 28(5), 486–510, 2023. [Link] [IZA No. 16092]
Economic Implications of Chinese Diagnosis-Related Group-Based Payment Systems for Critically Ill Patients in ICUs (with Zhaolin Meng, Yanan Ma, Suhang Song, Ye Li, Dan Wang, Yafei Si, Ruochen Zhang, Hao Xue, Limei Jing and Huazhang Wu). Critical Care Medicine, 48(7):e565-e573, 2020 [Link]