Job Market Paper
Between 1995 and 2010, China's Three Gorges Dam uprooted more than one million people, resulting in the largest involuntary displacement from dam construction in history. This paper provides the first evidence of the causal impact of dam-induced inundation on migration and labor market outcomes by combining micro-level census and satellite data. Using a novel identification strategy and remote sensing techniques to capture exogenous variations in flooding intensity, I find that inundation has imposed large and enduring costs on the local economy. Rising water levels in the dam's reservoir displaced between 1.5 and 1.9 million people. Most migrants in partially flooded counties relocated to other towns or villages within the same county. Flooded counties saw a steep and persistent decline of employment by 30 to 50 percent in the manufacturing sector and capital-intensive occupations, and the effects were more harmful to residents that did not move. In the long run, the decline in manufacturing was partly offset by reallocating workers to agriculture. Industrial firms in flooded counties responded by downsizing the workforce and reducing compensation. Overall, these findings highlight the need for policy evaluations to carefully weigh the broad benefits of infrastructure against the concentrated costs to local communities.
with Johannes Urpelainen (Johns Hopkins SAIS) [Appendix]
Divisive facilities, such as wind turbines, are important for society but controversial because they often carry local costs. Elections are the most important means for citizens to express their views about divisive facilities, but the political science literature on the topic has little to say about their electoral effects. To fill the gap, we examine the case of wind turbine construction in the United States. Using the average wind potential to instrument for wind turbine installation, we show that between 2003 and 2012, wind power generated large electoral benefits for (pro-renewables) Democratic candidates: every 100 megawatts of wind capacity increased Democratic vote share in U.S. House elections by 2-3 percentage points. This electoral shift, in turn, has contributed to a pro-environmental shift in congressional roll call voting on the environment. While the result is inconsistent with naive models of retrospective voting, it accords with positive reinforcement theories on how policies endogenously create their political support.
with T. Carleton, M. Delgado, M. Greenstone, T. Houser, S. Hsiang, A. Hultgren, A. Jina, R. Kopp, K. McCuster, I. Nath, J. Rising, A. Rode, H. K. Seo, J. Simcock, A. Viaene, J. Yuan
Using subnational data from 41 countries, we develop an empirical model of the mortality-temperature relationship that allows us to estimate effects where no mortality data exist and to account for the benefits of adaptation to climate. Importantly, we develop a revealed preference approach that bounds adaptation costs, even though they cannot be directly observed. Using future climate simulations, we compute a median willingness-to-pay of $20 (moderate emissions scenario) to $39 (high emissions scenario) to avoid the excess mortality risk caused by a 1t increase in CO2 emissions (2015 USD, 3% discount rate). Allocating these costs to 24,378 political units, we find substantial heterogeneity.
How Much does Environmental Quality Matter? Evidence from a Contingent Valuation Survey in China
with Dong Guo and Anyi Wang (Columbia Earth Institute) Under Review
In many developing countries, rapid industrialization and urbanization are often accompanied by deteriorating environmental quality that imposes substantial health and productivity costs on the local population. But existing studies have generally found low marginal willingness to pay (WTP) for mitigating such damages. To understand the magnitude and determinants of WTP for environmental quality improvements, we conducted a large in-person survey in three Chinese cities in 2016 and collected comprehensive demographic, socioeconomic and stated preference information from more than 3,000 respondents. Combining this survey data with ground-level air quality monitor data, we use a contingent valuation method to find that exposure to air pollution is a significant determinant of the extent individuals are willing to pay for improvements in air and overall environmental quality.
with Johannes Urpelainen (Johns Hopkins SAIS) and Wolfram Schlenker (Columbia SIPA), Center on Global Energy Policy, November 2018
with Yao Lu (Columbia Sociology), Edward Elgar Publishing, December 2016