Advances in Causal Inference at the Intersection of Air Pollution and Health Outcomes

Dylan Brewer, Daniel Dench, and Laura Taylor
About This Project

Professor and School Chair Laura Taylor and assistant professors Dylan Brewer and Daniel Dench published "Advances in Causal Inference at the Intersection of Air Pollution and Health Outcomes" in the Annual Reviews of Resource Economics. Brewer writes:

In the article, we review the methodological contributions that economists have made to studying the health effects of air pollution. Most non-economist approaches to studying the health impacts of air pollution rely on correlations between pollutants and health outcomes. The economics field has contributed a new methodology that seeks to establish causal relationships that go beyond simple correlations. To do so, economists study natural experiments in air pollution that mimic a lab experiment, where  some people are randomly exposed to air pollution (the treatment group) and others are not exposed (the control group). 

For example, a traditional air pollution study would use data on the health of people who lived near a pollution source such as a coal-fired power plant. The problem with this study is that the power plant may have been built in a low-income location or low-income residents may have moved near the power plant to take advantage of lower rent or property prices, which may conflate the effect of income and air pollution on health. An economist may try to overcome this by studying households that are downwind compared to households that are upwind of the power plant during a given period. As long as the wind direction is random, this can help overcome confounding effects such as income.


This article provides an overview of the recent economics literature analyzing the effect of air pollution on health outcomes. We review the common approaches to measuring and modeling air pollution exposures and the epidemiological and biological literature on health outcomes that undergird federal air regulations in the United States. The article contrasts the methods used in the epidemiology literature with the causal inference framework used in economics. In particular, we review the common sources of estimation bias in epidemiological approaches that the economics literature has sought to overcome with research designs that take advantage of natural experiments. We review new promising research designs for estimating concentration-response functions and identify areas for further research.

Read more:

Discover more featured research from the School of Economics, and follow us on Facebook, Twitter, Instagram, and LinkedIn to keep up with our students, school news, and upcoming events.