Kaiyi Wen 温开一

We contribute to the literature on environmental justice by exploring the role of two relatively unexplored factors, redlining and kinship, in explaining the environmental pollution gap between marginalized and non-marginalized communities. Using restricted data on 23,000 survey respondents in the contiguous US, we measure onsite emissions, ambient road noise, and PM2.5 concentrations at the individual and community levels. We find that individuals identifying as Black are exposed to higher levels of all three pollutants as compared to White individuals. However, the results are sensitive to geographic scale and it matters whether we measure pollution at the local (i.e. individual) or neighborhood scale. Overall, our results suggest that while neighborhoods with higher shares of African Americans have no greater ambient pollution than relatively white neighborhoods, within these neighborhoods, individuals of color may still be disproportionately exposed to pollutants as compared to white individuals.

Presented at AERE 2024 Summer Conference; 2024 NAREA Annual Meeting and Workshop; SEA 94th Annual Meeting; *MEA 2025 Annual Conference; *WEAI 100th Annual Conference.

While federal regulation has delivered major improvements in air and water quality, sensory pollutants — such as traffic noise and artificial nighttime illumination — remain largely unregulated and understudied. These environmental stressors are pervasive in urban infrastructure and increasingly linked to mental health risks, yet little causal evidence exists on their impacts. We provide the first national-scale analysis of the effect of ambient roadway noise and nighttime light pollution on mental health. We link restricted-use survey microdata from approximately 14,000 respondents in the United States to high-resolution sensory exposures, and leverage natural variation in topography, temperature, and atmospheric conditions to isolate the causal effect of roadway noise and nighttime light on self-reported mental health. Both pollutants exert statistically and economically meaningful effects: road noise increases the likelihood of mild psychological distress by 11%, while light pollution increases it by 13%. These effects are partially mediated through reductions in sleep duration. Welfare losses are up to $25 billion annually.

Note: This paper is a unified and substantially revised version of my two earlier working papers, “Neglected Hazard: Mental Health and Roadway Noise” and “Too Shiny to Handle: Nighttime Light and Mental Health.” The current version integrates both projects into a single conceptual and empirical framework and contains significant new analyses and results.

Presented at AERE 2023 Summer Conference; 2023 HINTS Data Users Conference; *PSU AGSE Seminar; *Delhi School of Economics Seminar; *MEA 2024 Annual Conference; CES 2024 North America Conference; *Indian Statistical Institute Seminar; OSWEET Online Workshop (September 2024); 2024 Heartland Workshop; *SEA 94th Annual Meeting; Global GLO-JOPE Conference 2024; 2025 ASSA Annual Meeting; EEA 51st Annual Conference; AERE 2025 Summer Conference.

This paper estimates the causal impact of a major supply-side shock in the U.S. illicit opioid market. Beginning in 2012, Mexican drug trafficking organizations displaced Colombian suppliers and took control of white-powder heroin distribution in the Eastern U.S. This transition raised the mean and variance of heroin purity and facilitated widespread fentanyl adulteration. Using 2004–2019 discharge-level data from the National Inpatient Sample and an event-study design, we show that this shift led to a 73 percent (0.11 percentage point) increase in heroin overdose inpatient admissions in affected regions. Placebo tests using alcohol poisoning and cocaine show no comparable effects, while spillovers to fentanyl- and methadone-related admissions are positive but imprecise. The effects are concentrated among Medicaid-covered and self-pay patients, indicating substantial public sector exposure. We estimate direct medical costs of approximately $4 billion, but the total welfare loss is far larger once accounting for excess mortality, reduced labor force participation, and intergenerational deficits in human capital.

Environmental injustice can be explained by two mechanisms: the disproportionate siting of polluting industries in low-income and minority communities, and demographic changes following the establishment of toxic sites. This paper investigates the siting decision of fossil fuel power plants and the subsequent demographic changes due to the siting in the context of environmental injustice. Using data at census tract level in the United States from 2000 and 2019, we find that fossil fuel power plants are more likely to be located in areas with higher minority populations, which leads to an average increase in the local minority ratio by 1.2%.

Presented at *2023 ASSA Annual Meeting.

We provide quasi-experimental estimates of the impact of coal and natural gas power plant retirements on the mental health of local residents in the United States. Combining data on power plant retirements and restricted mental health data, we employ a difference-in-differences approach and find that coal-fired power plant retirements have a significant negative impact on mental health, while natural gas retirements have a positive effect. We explore potential mechanisms and find evidence suggesting that economic impacts and local amenity improvements drive these divergent effects. Our findings highlight the importance of considering mental health implications in energy transition policies and strategies.

Presented at *2025 ASSA Annual Meeting; *AERE 2025 Summer Conference.

This paper examines whether individuals' behavioral responses to cigarette taxes shifted during the Great Recession. Using repeated cross-sectional data from the Behavioral Risk Factor Surveillance System (BRFSS) between 2003 and 2009, we estimate linear probability models to assess the association between cigarette tax increases and smoking cessation behaviors before, during, and after the recession. Our empirical strategy exploits variation in state-level cigarette taxes and their interaction with a recession-period indicator, while controlling for demographic characteristics, state and year fixed effects, and state unemployment insurance generosity. Contrary to standard economic predictions, we find that individuals were less responsive to cigarette tax increases during the recession, even though overall quit attempts modestly increased. Heterogeneity analyses reveal that this diminished responsiveness was particularly evident among lower-income individuals and those without health insurance — groups more vulnerable to financial and psychological stress. These results suggest that economic downturns may constrain individuals' capacity to respond to price-based health policies, with important implications for the timing, equity, and effectiveness of tobacco control strategies.