Efficiency and Equity Impacts of Urban Transportation Policies with Equilibrium Sorting (with Panle Barwick, Shanjun Li, Andrew Waxman, and Jing Wu) American Economic Review, 114 (10): 3161–3205.
We estimate an equilibrium sorting model of housing location and commuting mode choice with endogenous traffic congestion to evaluate the efficiency and equity impacts of a menu of urban transportation policies. Leveraging fine-scale data from household travel diaries and housing transaction data identifying residents’ home and work locations in Beijing, we recover structural estimates with rich preference heterogeneity over both travel mode and residential location decisions. Counterfactual simulations demonstrate that even when different policies reduce congestion to the same degree, their impacts on residential sorting and social welfare differ drastically. First, driving restrictions create large distortions in travel choices and are welfare reducing. Second, distance-based congestion pricing reduces the spatial separation between residences and workplaces and improves welfare for all households when it is accompanied by revenue recycling. Third, sorting undermines the congestion reduction under driving restrictions and subway expansion but strengthens it under congestion pricing. Fourth, the combination of congestion pricing and subway expansion delivers the greatest congestion relief and efficiency gains. It can also be self-financed, with the cost of subway expansion fully covered by congestion pricing revenue. Finally, eliminating preference heterogeneity, household sorting, or endogenous congestion significantly biases the welfare estimates and changes the relative welfare rankings of the policies.
Welfare Effects of Resale Price Maintenance: Evidence from the Chinese Pharmaceutical Industry (JMP). Revise and resubmit at Journal of Political Economy
This paper studies how resale price maintenance (RPM), a vertical practice that allows upstream manufacturers to directly control downstream retail prices, affects welfare. This effect is theoretically ambiguous because RPM can increase consumer welfare by eliminating double markups, but decrease it by facilitating price coordination across competing retailers. I provide empirical evidence of these effects by examining a high-profile antitrust case involving a pharmaceutical firm that was found to practice RPM and ruled to stop. Difference-in-differences estimates using a novel retail pharmacy-level dataset suggest the net effect of RPM decreases retail prices and is pro-competitive in this setting. The results also suggest that RPM reduces retailer markups and suppresses price dispersion across retailers. These findings are consistent with manufacturers’ incentives of using RPM to both eliminate the double markup and coordinate the retail price. I build and estimate a structural model to disentangle these incentives. The model confirms that RPM is overall welfare-improving in this setting. However, the consumer surplus gain from RPM would have been 77% higher absent of price coordination incentives. In addition, I show that under different market conditions RPM can lead to anti-competitive outcomes.
A Double Dose of Reform: Insurance and Centralized Negotiation in Drug Markets. (with Panle Barwick and Ashley Swanson) Featured in NBER Digest. Revise and resubmit at American Economic Review
Making innovative drugs affordable and accessible is a pressing global challenge. Centralized negotiation is an increasingly popular policy solution, but it remains understudied despite wide variation in implementation. This paper studies China’s ongoing NRDL Reform, which combines centralized drug price negotiation with expanded insurance coverage. The reform reduced retail prices by 48% and out-of-pocket costs by 80%, and increased drug utilization by 350%. At the same time, the insurance design was regressive, and 25% of negotiations failed. Focusing on cancer drugs, we estimate a flexible demand and supply model that features heterogeneous households, bargaining with potential breakdowns, and a government objective function that depends on consumer surplus and insurance spending. We estimate that including innovative cancer drugs in the NRDL generated RMB 40 billion ($5.6 billion) in annual consumer surplus gains and increased survival by 900,000 life-years among Chinese cancer patients each year. Among the counterfactual policies we examined, centralized market-access negotiation with an optimal coinsurance schedule raises social surplus by 19% relative to the observed policy and achieves 90% of the social surplus of an efficient benchmark.
Supporting Content Creators on Two-Sided Markets: Experimental Evidence from a Short-Form Video Platform. Revise and resubmit at Management Science
A few top content creators capture most of the impressions on digital platforms, discouraging grassroots users from creating new content, and thereby threatening the platform ecology. This concentration presents a dilemma to the platform: whether a platform should capitalize on established content creators’ popular content in the short run or promote content creation from amateurs in the long run. To quantify this tradeoff, I study a two-sided experiment by a short-form video platform that exogenously increases the impressions of treated amateur-generated content to treated viewers. Although viewer usage time decreased in the short run, the program successfully fostered the production of higher-quality and more diverse content from amateur creators. Overall, incentivizing amateur content creation offered net benefits to platforms in three months.
Range Anxiety (with Panle Barwick and Shanjun Li).
Range anxiety, the fear of depleting battery before reaching a charging station, is often cited as a major barrier to electric vehicle (EV) adoption, yet there has been limited formal economic analysis to quantify its importance and understand the policy implications. We develop a continuous-time dynamic model of EV usage and charging decisions to quantify range anxiety as the utility loss from feasible yet unrealized trips due to perceived range constraints. Using high-frequency data of 188,000 EV trips and 30,000 charging events among 8,000 EVs in Shanghai, we recover model parameters governing consumer driving and charging decisions. The estimates imply that, across EV models with varying driving ranges, average range anxiety was about $1,900 in 2021 but declined to $1,200 in 2024, driven by improvements in charging infrastructure and, especially, increases in driving range. Policy simulations underscore the importance of coordinating investments in battery capacity and charging infrastructure to address range anxiety: relative to socially optimal levels, Shanghai’s EV market has under-invested in driving range while over-investing in charging infrastructure.
From Pumps to Plugs: Value of Time and Charging Policies. (with Takeaki Sunada)
Designing EV charging policies requires understanding how drivers trade off charging speed, price, and convenience. Using high-frequency vehicle telemetry data from Shanghai, we estimate a random-coefficient discrete-choice model of charging demand and provide field-based estimates of drivers’ value of time (VOT) in charging. The mean VOT is 27% of the average wage, implying lifetime charging time costs of 1.7% of the vehicle price. Counterfactuals show that Level-III charging speed reduces driver time costs by 70% and explains 4.6% of new vehicle adoptions. We evaluate how improving vehicle charging speed, station power, and station density affect driver welfare.
From Free Rider to Innovator: The Rise of China’s Drug Development. (with Panle Barwick and Hongyuan Xia)
Can a developing country escape the pharmaceutical ``free-rider trap’’? We study China’s rapid transition from a producer of generic drugs to a major source of frontier pharmaceutical innovation. Compiling a harmonized database of clinical trials, scientific publications, drug sales, and policy reforms, we document a structural break beginning in 2016: China’s clinical trial volume surged by 86.1\% relative to the U.S., with comparable gains in {trial novelty}, concentrated among domestic firms. We examine a comprehensive list of potential demand- and supply-side drivers, including the National Reimbursement Drug List (NRDL) reform, knowledge accumulation, talent flows, regulatory streamlining, Made in China 2025, international harmonization, and other state-led initiatives. China’s rise in drug development was primarily driven by NRDL, where reimbursed disease categories experienced significantly greater trial growth, but never-covered categories exhibited no change. A decomposition exercise attributes 43\% of the trial surge in Oncology to NRDL, versus 24\% for knowledge accumulation and talent flows, with other factors playing a modest role. Back-of-the-envelope welfare calculations suggest that the dynamic gains from induced innovation are three times the static gains from expanded patient access. Our findings demonstrate that well-designed demand-side policy can simultaneously improve affordability and stimulate frontier R\&D, offering a pathway for developing nations to contribute to global pharmaceutical innovation.
(Click on the title to view abstracts.)
In the market for medical goods and services, the intense involvement of the public sector creates a possible linkage between public-sector policies and private-sector outcomes. This study investigates the external effects of a centralized procurement auction policy on generic drugs in China, which creates a price shock (50% decrease on average) in the public sector. Leveraging the regional and timing variations, I find that the pharmacies’ retail price response is much smaller (10%), indicating strong market friction. I build a structural model to quantity the welfare and provide evidence of consumer inertia and transportation costs as the main mechanisms that explain market friction. Firstly, patients that got diagnosed recently are more likely to switch, indicating strong inertia. Secondly, the price response is more significant for pharmacies closer to hospitals, consistent with high transportation costs.
In the year 2016-2018, the Ministry of Public Security of China implemented a new policy to replace the electric vehicles’ (EV) plates with green ones, making EVs’ plates distinguishable from gasoline cars. We use the differences-in-differences method, leveraging the staggered implementation of the policy to identify the treatment effect. We find that this ”nudge” policy boosts the local sales of EVs by 24%, saving the government 2.5 billion RMB in cash subsidy. This effect is robust to province boundary design and heterogeneous across different EV brands and city demographics. We provide additional thoughts on the mechanism.
This paper examines the effects of social pressure on an individual’s behavior. I investigate the policy of a library that publicizes names of borrowers who fail to return books on time to see how such “shame tactics” induce the more timely returns. First, I use a difference-in-differences method to identify the impact of social pressure using historical policy changes as a quasi-experiment. Then a randomized controlled trial (RCT) is conducted by sending emails with different contents to students. The results indicate that: (1) social pressure increases the on-time return rate by 5 percentage points, comparable to the effects of reminders, but weaker than a small fine (1 RMB a day); (2) the impact of social pressure is heterogeneous over different groups of individuals, and is especially strong for faculty, students with wider social connections, and individuals who would have very likely returned the books on time. In practice, this paper suggests an alternative policy tool that facilitates task completion.