Working Paper: NBER ID: w23524
Authors: Alan I. Barreca; Matthew Neidell; Nicholas J. Sanders
Abstract: Though over 90 percent of benefits from environmental quality improvements are attributed to long-term exposure, nearly all quasi-experimental evidence on the effects of pollution on health exploits changes in short-term exposure. Quantifying long-run exposure impacts requires a lasting, exogenous change in ambient pollution. Even if the initial change in pollution is exogenous, the long-run nature allows more time for economic agents to respond to changes in pollution, resulting in endogenous pollution exposure. We estimate the effects of long-run pollution exposure on mortality among adults by exploiting the United States Acid Rain Program (ARP) as a natural experiment. The ARP, which regulated emissions from coal power plants, created a permanent change in pollution across vast distances, enabling us to define broad treatment areas to subsume many potential confounding effects. We use a difference-indifferences design, comparing changes in mortality over time in counties “near” regulated plants to changes in mortality in similar counties “far” from the plants. We find relative mortality in treatment counties decreased after the introduction of the ARP, with mortality improvements growing steadily over time in both economic and statistical significance. The ARP had no significant effect on residential sorting or employment, helping rule out selection or economic mechanisms. Analysis by cause of death supports the role of fine particulate matter as the relevant pollutant.
Keywords: Pollution; Mortality; Acid Rain Program; Health Economics
JEL Codes: I10; Q51; Q53
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| ARP (Y60) | mortality in treatment counties (I12) |
| SO2 emissions regulation (L98) | PM2.5 exposure (I14) |
| PM2.5 exposure (I14) | mortality (I12) |
| ARP (Y60) | labor market outcomes (J48) |
| ARP (Y60) | residential sorting (R23) |
| mortality in treatment counties (I12) | mortality improvements over time (C41) |