I am a Ph.D. candidate in Economics at Pennsylvania State University, specializing in industrial organization, causal inference, and applied microeconomics. My work combines causal inference, structural modeling, and large-scale data analysis to understand consumer and firm behavior and support product and business decision-making.
Most recently, I worked as an Economist Intern at Amazon, where I used household-level data and quantitative modeling to study subscription behavior and evaluate product strategy.
I am on the 2026–2027 job market and am seeking industry roles.
Before Penn State, I received a B.A. in Economics from Peking University.
Economist Intern, May–August 2026
I worked on subscription and family-plan strategy, using household-level data to understand how customers make joint subscription decisions and how benefit sharing affects product outcomes. I extended an existing demand framework from individual to household-level choice and built scalable data pipelines to create model-ready household panels.
My work combined discrete-choice modeling, multi-period counterfactual simulation, model validation, and large-scale data engineering using Python, PySpark, SQL, and AWS. I evaluated alternative product scenarios and quantified tradeoffs among paid subscriptions, customer access, and engagement, then translated the results into product and business implications for economists, scientists, and engineering stakeholders.
Job Market Paper
Abstract: Intellectual property policy affects innovation by shaping the returns to developing and trading inventions. This paper studies how earlier patent disclosure changes patent trade and firms' research incentives in the U.S. pharmaceutical industry. The American Inventors Protection Act (AIPA) of 1999 shortened the period during which patent applications could remain secret. Linking patent records, ownership transfers, and firm financial data, the empirical analysis documents broader participation in patent trade after AIPA and a more positive relationship between patent quality and transfer probability. Among transferred patents, the time from application to transfer falls more in technology groups with longer disclosure delays before AIPA. To examine the implications for innovation, we develop and estimate a model of R&D investment and patent trade using data on public pharmaceutical firms, where earlier disclosure improves matching between inventors and potential patent buyers while reducing inventors' private returns to retaining patents. Counterfactual analysis shows that AIPA lowers aggregate R&D spending but raises patent output when patent trade is available, because opportunities to trade patents support greater research investment among firms with lower initial R&D investment or less capacity to commercialize patents themselves.
Abstract: This paper studies the effects of New York City’s recent short-term rental (STR) regulation using a comprehensive monthly panel of Airbnb listings. The policy requires STR hosts to obtain a city registration and imposes additional operational restrictions on both hosts and guests during active stays. I document that the regulation induced a large share of hosts to transition into long-term rental markets or to exit the platform entirely. In the aftermath of the policy, listing prices rose sharply and the dispersion of listing quality shifted. To interpret these patterns, I develop a dynamic model in which hosts make forward-looking pricing, entry, and exit decisions. I use the model to quantify the regulatory cost burden faced by hosts before and after the policy change. The estimates imply that the regulation substantially reduces aggregate welfare, with total welfare falling by 83.94%.
with Yu Gao, Ying Lei, Shuo Liu, Juanjuan Meng, and Xi Weng
Abstract: We conduct a large-scale survey experiment across six culturally diverse countries (China, Germany, Japan, Saudi Arabia, Singapore and the United States) to test whether information about privacy-enhancing technologies (PETs), such as encryption and federated learning, reduces concerns about personal data collection and increases acceptance of facial recognition. We also examine heterogeneity by digital literacy and privacy motivation. We find that PETs information lowers privacy concerns in Germany, the United States, and China, particularly among individuals with lower digital literacy and those motivated by instrumental rather than intrinsic considerations. However, it has no effects on acceptance of facial recognition. These findings highlight both the reach and limits of technical assurances, and underscore the relevance of distinguishing between instrumental and intrinsic privacy motivations in shaping public attitudes toward emerging technologies.
Fields & Methods: Causal Inference · Applied Microeconomics · Structural Modeling · Industrial Organization
Programming & Data: Python, SQL, PySpark, R, Stata, MATLAB
Cloud & Workflow: AWS (EMR, S3), Airflow