To be presented at
Econometric Society World Congress 2025
19th North American Meeting of the Urban Economics Association
GSE-OSIPP-ISER Joint Conference in Economics 2025 (University of Osaka)
Abstract: Yardstick competition between local governments can generate internal trade barriers. I exploit plausibly exogenous variation in competition intensity arising from age differentials between Chinese city leaders to show that intensified competition reduces city-level trade flows and export values. Analysis of bilateral trade data indicates these effects stem from policies at the origin, creating export barriers that fragment domestic markets. The induced fragmentation disrupts firm-level supply chains, thereby lowering both labor productivity and revenue. Back-of-the-envelope calculations suggest a one-standard-deviation increase in competition intensity causes a 0.56% GDP loss through reduced trade in intermediate goods. Politicians implement internal trade barriers via three policy tools: politicians facing stronger competition strategically shift industrial subsidies and land allocation toward non-tradable real estate sectors while simultaneously cutting investment in transportation infrastructure. The findings demonstrate how institutional design can endogenously generate barriers to economic integration, with implications for other decentralized nations.
[Excellent Paper of the 17th Applied Econometrics Conference] Revise & Resubmit at Economic Development & Cultural Change
Abstract: Government subsidies can boost firm performance and growth, but political connections, especially those based on shared hometowns, can lead to politician-firm corruption and distorted subsidy distribution. This study examines how government subsidies given to firms in politicians' hometowns are affected by anti-corruption efforts. Using a Difference-in-Differences approach and China's 2013 anti-corruption inspections as the exogenous shock, I find that the inspection reduces subsidies by 22.6% for firms in cities with a native-born leader. Also, a novel measure of hometown favoritism—quantifying leaders born in each city but serving elsewhere—reveals that each additional leader a city produced before the inspection results in a subsequent 2.2% subsidy cut. The back-of-the-envelope calculation suggests that the pre-existing corrupt subsidies linked to city leaders' hometowns are estimated to cost about $2.24 billion, or 0.02% of China's GDP. The anti-corruption inspection decreases vaguely described subsidies, reduces subsidies generated from firms' corrupt spending, and increases government public spending in affected politicians' hometowns. This increase in public spending is likely due to the crowding-out effect of previous corrupt subsidies. Overall, these findings suggest that without anticorruption efforts, the politician-firm quid pro quo can distort the allocation of subsidies.
3.Covid-19, Migrant, and Automation (with Yalan Li)
Abstract: This paper examines the relationship between labor mobility disruptions and automation adoption using the COVID-19 pandemic as a quasi-natural experiment in China. We document two key stylized facts: COVID-19 restrictions significantly reduced internal migration flows, particularly affecting low-skilled rural-to-urban workers, while more industrialized regions experienced disproportionately larger increases in automation intensity. Exploiting spatial and temporal variation in pandemic-induced migration restrictions across Chinese prefectures, we establish a causal link between labor supply disruptions and regional automation investments using pre-pandemic industrial structure and region-specific policy stringency measures, combined with data on International Federation of Robotics (IFR) robot installations and Chinese regional automation patent applications. To interpret these findings, we develop a spatial general equilibrium model featuring heterogeneous regions, endogenous migration decisions, and firms' optimal choice between labor and automation capital, incorporating agglomeration effects, regional productivity differences, and adjustment costs. Calibrated to pre-pandemic Chinese data, the model successfully replicates our reduced-form findings. Counterfactual simulations reveal that without migration disruption, automation adoption would have been 17% lower in highly industrialized regions, suggesting substantial pandemic-induced technological acceleration. Additional counterfactuals demonstrate that regions with greater initial automation capacity experienced smaller welfare losses, highlighting automation as both substitute for and complement to labor mobility in spatial equilibrium. Our results illuminate how temporary shocks generate persistent effects on regional development patterns and technological change.
4.Political Tenure, Land Reallocation, and Retreat from Trade
Abstract: I examine how the tenure of local political leaders affects land markets, fiscal capacity, and economic geography in China. Using a panel dataset linking city politicians to land transactions, fiscal accounts, and trade flows, I document three stylized facts: industrial land is initially priced substantially below residential land by local politicians, but this gap narrows as politicians' tenure lengthens; longer tenure correlates with systematic reallocation of land from industrial to residential use, reducing support for tradable goods production; and central government fiscal transfers decline with political tenure, increasing local governments' dependence on residential land sales revenue. To interpret these findings, I develop a multi-region spatial general equilibrium model with trade and costly migration in which local governments facing tenure-dependent fiscal transfers choose optimal land allocation between industrial and residential uses. Industrial land enhances tradable goods productivity and reduces migration costs, while residential land generates fiscal rents but only supports nontradable production. When calibrated to match city-level data on land prices, land-use shares, and tenure-trade elasticities, the model attributes approximately 30% of observed intercity trade decline and 25% of export contraction to tenure-driven land price convergence and reallocation effects. Counterfactual simulations show that policies stabilizing fiscal transfers or fixing industrial land prices would substantially mitigate these negative effects, increasing aggregate welfare by roughly 0.8%, with gains concentrated in more industrialized cities.
Abstract: This paper documents heterogeneous effects of agricultural price shocks on child labor and human capital in rural China, driven by crop skill intensity. Using a shift-share design interacting global price fluctuations with baseline local cropping patterns, we classify crops as labor- or skill-intensive via farmer education data. Price increases for labor-intensive crops significantly reduce middle school enrollment and increase child labor, while skill-intensive crop price increases promote higher education. Farm- and individual-level data corroborate these mechanisms via income, land use, and educational spending responses. Our findings show that these price dynamics yield lasting gender disparities in human capital, with crop-specific shocks imposing differential costs on girls' education and benefits for boys. This study illustrates a crucial mechanism through which trade shocks can perpetuate inequality and influence development trajectories.
Working in progress:
The Effects of International Agricultural Price Shocks on Internal Migration and Crime in China