Has Air Pollution Driven Skilled Workers Away from Chinese Cities?

This study investigates how pollution-induced migration affects productivity in China, focusing on whether skilled workers disproportionately leave polluted cities. Using data from China’s 2015 Population Census, labor surveys, and pollution metrics (e.g., satellite PM2.5 levels, wind patterns, thermal inversions), the study finds that a 10% increase in PM2.5 leads to a 1.4 percentage point rise in skilled worker emigration, compared to 0.61 for unskilled workers. This migration raises skill premiums in polluted cities and reduces national productivity. Counterfactual simulations suggest that reducing pollution could increase productivity by up to 12%, nearly matching the direct health benefits of cleaner air.

Full Citation and Link to Article

Khanna, Gaurav, Wenquan Liang, Ahmed Mushfiq Mobarak, and Ran Song, "The Productivity Consequences of Pollution-Induced Migration in China". AMERICAN ECONOMIC JOURNAL: APPLIED ECONOMICS (FORTHCOMING)

https://www.aeaweb.org/articles?id=10.1257/app.20220655&&from=f

Extended Summary

Central Research Question

This study examines the relationship between air pollution and labor migration in China, focusing on how pollution-induced migration affects productivity. The key question it addresses is: How does air pollution impact the geographic distribution of skilled and unskilled workers, and what are the productivity consequences of this re-sorting? The study investigates whether skilled workers are more likely to leave polluted cities than unskilled workers and quantifies the resulting impact on aggregate productivity, wages, and economic output. By using a spatial equilibrium model, it estimates how reducing pollution could influence productivity and overall economic welfare.

Previous Literature

The research builds on multiple strands of literature. One line of work focuses on the economic consequences of pollution, particularly its effects on health, productivity, and human capital accumulation. Studies have shown that pollution exposure reduces cognitive function, work efficiency, and life expectancy, providing clear economic incentives for individuals to relocate away from polluted areas.

A second body of literature examines internal migration and labor market dynamics, particularly in China, where the hukou system (a household registration policy) limits mobility for unskilled workers while offering more flexibility to skilled workers. Previous research suggests that migration restrictions exacerbate regional productivity gaps by preventing efficient labor allocation across cities.

Finally, the study contributes to urban and spatial economics by linking pollution to regional wage differentials and labor sorting. While prior work has documented that pollution affects migration decisions, this study quantifies the aggregate productivity effects of these migration patterns using a spatial general equilibrium framework. The authors also integrate empirical strategies from environmental economics, labor economics, and urban economics to provide robust causal estimates of pollution-induced migration.

Data

The study relies on multiple datasets to examine the effects of pollution on migration and productivity. The primary data sources include:

2015 China Population Census: This dataset provides detailed migration patterns, educational attainment, and hukou status, allowing the researchers to track individual relocation decisions.

China Laborforce Dynamics Survey (CLDS, 2016): This longitudinal panel tracks individual workers’ employment and migration histories, providing insights into mobility decisions over time.

Pollution Data: The study uses satellite-based PM2.5 air pollution measures from the Global Annual PM2.5 Grids dataset, which offers consistent pollution readings across Chinese cities from 1998 to 2015.

Meteorological Data: The researchers use temperature inversion data from MERRA-2, a NASA dataset that measures atmospheric temperature variations that trap pollutants in certain regions.

Economic and Industrial Data: The study incorporates city-level GDP, employment, and industrial composition statistics from China’s City Statistical Yearbooks and trade data from the China Customs Database.

Power Plant Data: Information on the locations, coal consumption, and output of large-scale power plants is used to instrument for exogenous variation in pollution levels.

By combining these diverse data sources, the study provides a comprehensive view of how pollution influences migration and economic outcomes.

Methods

The authors use a spatial general equilibrium model to quantify the effects of pollution-induced migration on productivity. This model accounts for wage determination, skill-biased labor mobility, migration costs, and the interaction between pollution, wages, and housing prices.

To establish causality, the study employs two instrumental variable strategies:

Wind Direction and Coal-Fired Power Plants: Cities located downwind of large coal-fired power plants experience higher pollution levels. The researchers use wind patterns and the locations of distant coal plants (100–500 km away) to instrument for exogenous variation in air quality.

Thermal Inversions: Temperature inversions trap pollution near the surface, creating natural variation in pollution levels. The study uses the frequency and strength of thermal inversions to instrument for pollution exposure across cities.

The researchers estimate migration responses using individual-level regression models where the dependent variable is whether an individual moved out of their hukou city. The main independent variable is log PM2.5 concentration, instrumented using the two exogenous pollution shocks.

To measure productivity effects, the study estimates:

Labor Demand Elasticities (σE): This determines how wage changes respond to shifts in skilled vs. unskilled labor supply. The authors exploit variation in pollution levels to identify the relationship between migration and skill premiums.

Labor Supply Elasticities (ηs, γs): The model estimates how migration decisions respond to pollution and wages, allowing the authors to quantify the extent of compensating differentials for air pollution.

Counterfactual Simulations: The researchers use the estimated model to predict how changes in pollution policies (e.g., reducing PM2.5 levels by 50%) would affect migration, wages, and overall economic output.

Findings/Size Effects

The study finds strong evidence that air pollution significantly influences migration decisions, particularly among skilled workers. The key results include:

Migration Effects: A 10% increase in PM2.5 concentration leads to a 1.4 percentage point increase in skilled worker emigration, compared to 0.61 percentage points for unskilled workers. This indicates that skilled workers are more sensitive to pollution, likely due to their greater financial and institutional flexibility.

Wage Effects: The relative scarcity of skilled workers in polluted cities raises the skill premium in those locations. Cities that experience skilled out-migration see a 14.5% increase in the returns to skill. This means that firms in polluted cities must offer higher wages to retain educated workers.

Productivity Effects: Counterfactual simulations show that reducing pollution increases productivity by 12%, largely due to improved labor sorting. This effect is comparable to the direct health benefits of cleaner air, suggesting that labor reallocation plays a crucial role in determining the economic impact of pollution.

Hukou Policy Effects: The study finds that mobility restrictions exacerbate productivity losses by limiting unskilled worker movement. Skilled workers are more able to relocate due to hukou exemptions, while unskilled workers often remain in polluted, low-wage cities, deepening regional disparities.

Overall, the findings highlight how pollution alters labor markets by redistributing skilled workers away from polluted but productive cities, reducing aggregate economic output.

Conclusion

This research provides a novel perspective on the economic costs of pollution by linking air quality to migration and productivity. The study shows that pollution not only affects health but also disrupts labor markets by driving skilled workers away from high-productivity regions. The authors quantify these effects using a spatial general equilibrium model, revealing that the productivity consequences of pollution-induced migration are as large as the direct health effects of pollution.

Policy simulations suggest that targeted pollution reductions—such as relocating upwind coal plants or setting city-specific emissions caps—could significantly improve national productivity. Additionally, relaxing hukou mobility restrictions would allow unskilled workers to relocate more freely, mitigating some of the welfare losses associated with pollution-induced labor sorting.

The broader implications extend beyond China. In many urban centers worldwide, pollution affects residential and occupational choices, influencing regional productivity and wage disparities. This study underscores the importance of integrating labor mobility considerations into environmental policy discussions, as addressing pollution could yield substantial economic benefits beyond traditional health improvements.