Working papers
Welfare Effects of Subsidizing Public Transit for Older Adults with Kanghyock Koh, Suk Joon Son, and Jung-suk Lee
This study examines the effects of subsidizing public transit for seniors, focusing on impacts through externalities and internalities. Using novel cell-phone mobility data and leveraging the eligibility threshold for free subway rides at age 65 in Korea, we find that the free subway policy significantly increases subway usage during both peak and off-peak hours but has only a minimal impact on reducing car usage. The current policy appears highly regressive, benefiting seniors in affluent neighborhoods the most. Additionally, we observe an overall increase in transportation usage, leading to more visits to locations that may provide health benefits.
We estimate a transportation mode choice model and a location visit prediction model to simulate counterfactual outcomes for alternative policies currently under discussion. These include a free subway policy limited to off-peak hours to reduce subway congestion and a budget-equivalent discount policy that subsidizes both buses and subways to improve access equity. We characterize an optimal policy that considers congestion, redistribution preferences across generations and income groups, and the paternalistic goal of enhancing seniors' health. Our findings suggest that efficiency costs and redistribution preferences are the most critical factors for the optimal policy, while health benefits—often emphasized by policymakers—should not be the primary factor in policy design.
Externality of Driving Luxury Vehicles and Optimal Taxation with Jaewon Lee, and Suk Joon Son
Under tort law, where the at-fault driver is responsible for covering the repair costs of another party, driving a luxury vehicle with higher repair costs creates a negative externality. A Pigouvian tax on luxury vehicles, or a vehicle-value-based premium, can help internalize this externality. Using novel micro-level data on automobile sales and repair costs, and leveraging the introduction of a luxury vehicle tax in British Columbia, we demonstrate that a typical luxury vehicle generates an externality of $0.10 per dollar of vehicle cost. We estimate a structural model showing that the optimal tax would increase welfare by 0.8% of the British Columbia automobile industry's value, extrapolating to an $8 billion increase in the U.S. automobile market.
Propensity to Consume Food Out of SNAP and its Welfare Implications (Revise & Resubmit, Journal of Public Economics)
Using new consumer-panel data across multiple grocery retailers, I estimate that the marginal propensity to consume food out of SNAP benefit dollars is 47.8 cents. This is lower than the 59 cents reported by Hastings and Shapiro (2018), who used data from a single retailer. I find that providing in-kind benefits rather than cash results in an increase of 921 calories per day in total food purchased, while dietary quality remains unaffected. Assuming in-kind transfers induce a choice mistake due to mental accounting, I estimate an efficiency loss of 12 cents per dollar of in-kind transfers compared to cash.
Insurance with Multiple Agents with Jason Abaluck and Oren Sarig
In most insurance contracts, multiple parties jointly determine expenditures: patients and doctors in health insurance, drivers and mechanics in auto insurance, or workers and employers in unemployment insurance among others. In such settings, we analyze how an insurer or social planner divides cost-sharing between risk-averse consumers and risk-neutral firms. Insurers use exclusively consumer cost-sharing only in the limit where firms are perfectly competitive. In that limit, firm cost-sharing is ineffectual since firms must maximize consumer utility. With even an ε amount of market power, insurers use firm cost-sharing to control costs for high spending consumers. When firms have substantial market power, consumer cost-sharing is necessary only to control "extensive margin'' moral hazard beyond the direct reach of firms, such as whether to visit the doctor at all. The model explains why, in healthcare, increasing provider concentration has been accompanied by a shift from coinsurances (consumer cost-sharing) to prior authorization (firm cost-sharing). We show in the setting of prescription drug insurance that prior authorization is far more impactful than coinsurances per dollar, as the model predicts with high provider concentration.
Work in Progress
Revisiting SNAP’s Impact on Birth Outcomes with Mohit Agrawal, Kinsey Dinan and Kacie Seil
(Received data from NYC DSS and DOHMH, grant supported by Policy Impacts)
Optimal Insurance for GLP-1 for Overconfident Consumers with Michael Darden and Yaa Akosa Antwi
(First pilot survey completed)