Research
Publication
- “Industrialization and Pollution: The Long-Term Impact of Early-Life Exposure on Human Capital Formation.” January 2025, Journal of Public Economics. Manuscript
Abstract
Air quality in developing countries is often much worse than in developed economies, yet evidence on the long-term human capital effects of air pollution in these settings is limited. This paper uses a cohort difference-in-differences approach to examine the impact of early-life exposure to air pollution during China's 1950s industrialization on human capital formation. It assumes that economic opportunities linked to industrial plants impact upwind and downwind counties similarly within a 30-mile radius. The results indicate that moving from the 25th to 75th percentile of exposure reduces children's education by approximately 0.11 years. This effect size is notably larger than the impacts of three other factors affecting educational attainment in both China and the United States.
Working Papers
- “Powering the Future: The Long-Term Human Capital Effects of Rural Electrification”
Abstract
In low- and middle-income countries, electricity often suffers from coexisting affordability, reliability, and stability challenges, yet empirical work jointly examining these dimensions remains limited. This paper studies the long-run human capital effects of a rural electrification program implemented in China in the late 1990s that simultaneously improved all three dimensions. Using a cohort difference-in-differences design with treatment intensity measured by county-level electricity price reductions, I find that exposure during middle childhood (ages 6--11) leads to substantial long-run gains in human capital. In contrast, exposure only after middle childhood yields negative and statistically insignificant effects. Analysis of mechanisms shows that improved agricultural productivity is an important channel and that increased public investment in education may also play a role. Extended study hours due to improved lighting, however, are unlikely to be relevant. China's experience offers insights for current rural electrification efforts in developing countries. - “Environmental Regulation and Within-Firm Adjustment: Evidence from Chinese Steel,” with Feitao Jiang and Yingjun Su
Abstract
Firms can comply with environmental regulation by changing production processes or by abating pollution after it is generated; yet the production margin remains largely a black box. This paper opens it using stage-level production, environmental, and financial data for Chinese steel firms. We exploit within-province variation in regulatory pressure under the 2016–2020 Good-Air Days targets in a difference-indifferences design. Firms facing greater pressure rely less on the pollution-intensive upstream stages and invest more in abatement. A one-standard-deviation increase in target in-tensity reduces pig iron input per ton of crude steel by 2.4% of its mean, implying a 2.2-percentage-point rise in scrap’s share of metal inputs. Environmental investment rises by 49% of its mean, concentrated in air-pollution control. Pre-tax profit per ton does not fall. The production adjustment implies only a small reduction in reported SO2 emissions. Evaluations based solely on emissions or financial outcomes would overlook the production margin - “Deciding to Participate: The Impact of Air Pollution on Civic Engagement in China”
Abstract
Online engagement with government is increasingly common in the digital age, yet the factors driving such civic activity remain poorly understood. This paper uses an instrumental variable (IV) strategy to examine whether air pollution influences online engagement with the government, measured by message volume on an official platform in China. I find that a 10 ug/m3 increase in weekly average PM2.5 results in a 15.9% surge in messages. During periods of higher pollution, people are more likely to voice complaints, seek assistance, make inquiries, and offer suggestions. Two mechanisms are identified to drive this response: air pollution shifts the perceived payoffs of civic engagement, and it heightens public awareness of everyday problems. Sentiment analysis using large language models and dictionary-based tools shows that air pollution tends to worsen emotional well-being, although the effect is statistically insignificant in most cases. These findings carry practical implications: timely government responses to public concerns can help prevent grievances from escalating into more serious outcomes.
Selected Works in Progress
- “Text to Data: A Machine Learning Approach to Historical Chinese Documents,” with Wolfgang Keller, Carol H. Shiue, and Sen Yan