American Economic Review: Papers & Proceedings 2009, 99:2, 177–183 http://www.aeaweb.org/articles.php?doi=10.1257/aer.99.2.177

Air Pollution And HeAltH Around tHe World †

Fetal Exposures to Toxic Releases and Infant Health
By Janet Currie and Johannes F. Schmieder*
Every year, millions of pounds of toxic chemicals thought to be linked to developmental problems in fetuses are released into the air. Yet, we have only limited information about the health effects of these releases. A 1998 Environmental Protection Agency (EPA) review found that complete screening data about toxicity was available for only 7 percent of 3,000 chemicals released in large quantities in the US (US EPA 1998). Even for chemicals that have been studied, there is little information about how levels found in the environment affect human health. Laboratory data on toxicity may be of limited value, given that tests are typically conducted on animals, and do not take human behavior (such as staying inside on high pollution days) into account. Moreover, it is quite difficult to draw a relationship between a disease such as cancer and toxic exposures in a particular location, given that cancer develops over a long period, and people are mobile. In contrast, birth outcomes are likely to be highly affected by conditions during the brief interval of pregnancy (though of course they might also be influenced by factors affecting the mother before conception). Hence, infant health outcomes are an ideal place to look to see if existing environmental releases have detectable negative effects for human health. This study uses data from the EPA’s Toxics Release Inventory (TRI) matched to data from national Vital Statistics Natality and Mortality files to examine the effects of fetal exposure on health at birth and subsequent infant mortality. Exposure to toxic chemicals may be linked to many other characteristics of families and neighborhoods, and to swings in economic activity. In an effort to identify the effect of toxic exposures, we compare the estimated effects of chemicals that are thought to be developmental toxicants to those which are not known to have developmental effects. We also compare the effects of “fugitive” air releases to the effects of “stack” air releases. Emissions that go up a smoke stack are more likely to be treated in some fashion (e.g., with scrubbers), and travel farther than those that do not. Hence, they should be less likely to affect those in the immediate vicinity of the plant. Finally, we look at several of the most commonly known developmental toxicants separately.
I. Data and Methods

† Discussants: Heather Royer, Case Western Reserve University; Matthew Neidell, Columbia University; Lucas Davis, University of Michigan.

* Currie: Department of Economics, Columbia University, 420 W 118th St., New York, NY 10025 (e-mail: janet.currie@columbia.edu); Schmieder: Department of Economics, Columbia University, 420 W 118th St., New York, NY 10025 (e-mail: jfs2106@columbia.edu). The authors thank Alan Burke for excellent research assistance. We also thank Alan McCartland and participants in seminars at UC-Berkeley, Columbia University , and the International Health Economics Associations meetings in Copenhagen in June 2007 for helpful comments. This research was supported by funding from NIH R21 HD055613-01. We are solely responsible for all errors. 177

Information about pregnancy outcomes come from the Vital Statistics Natality data. They cover virtually all births and include information about characteristics of the mother, characteristics of the child, and health at birth. Information on infant deaths is taken from Vital Statistics Mortality files. We focus on birth weight, gestational age, and infant mortality in the first year of life, since there is considerable variation across counties in these outcome measures. Data on toxic releases come from the Environmental Protection Agency’s TRI, which was created by the Emergency