Individual Submission Summary
Share...

Direct link:

Air Pollution and Cardiovascular Mortality: Evidence from Thermal Inversions in the U.S.

Thursday, November 5, 8:30 to 10:00am, Property: Boston Marriott Copley Place, Floor: 4th Floor, Room: Vineyard

Abstract

This paper estimates the causal effect of fine particulate matter (PM2.5) on cardiovascular mortality in the United States using thermal inversions as an instrumental variable in a two-stage least squares framework. Using county month level data from 2010 to 2023 with 481,094 observations, results show that a one unit increase in PM2.5 increases the cardiovascular mortality rate by 3.1 to 4.6 percent. Cardiovascular disease is one of the leading causes of death in the U.S. with substantial economic costs including productivity loss. These findings show that moderate reductions in PM2.5 could reduce cardiovascular deaths which highlights the importance of accurate causal evidence for informing effective public health and environmental policies. A key contribution of this paper is a systematic evaluation of how thermal inversion construction choices affect the reliability of causal estimates. Prior studies mostly adopt a single construction without examining whether different methodological choices including the atmospheric data product, the pressure level selection, and the way thermal inversion is measured and aggregated affect instrument validity and the magnitude of estimates. I construct thermal inversion instruments under multiple specifications and evaluate each through first stage strength and plausibly exogenous analysis. Findings show that construction choices matter and affect both the strength of the first stage and the reliability of the estimates. The analysis also accounts for high terrain areas in the U.S. that prior studies did not address, which affects instrument validity in a national sample. The importance of the applied systematic approach is beyond this paper since thermal inversion is increasingly used as an instrument across a range of health and labor market outcomes. Thus, understanding how construction choices affect estimates has broader implications for the credibility of this identification strategy. Moreover, plausibly exogenous analysis confirms that the preferred specification remains robust to hypothetical modest violations of the exclusion restriction.

Author