« Abstract Submission Closed | Main | Complete List of Accepted Abstracts for 5th World Congress »
January 18, 2005
12th Annual Kenneth J. Arrow Award: Ken Chay and Michael Greenstone
Arrow Award Presentation at ASSA Meetings in Philadelphia on January 7, 2005
Presenter: Michael Grossman
In 1989 Ted Joyce, Fred Goldman, and I published a paper in the Journal of Urban Economics entitled “An Assessment of the Benefits of Air Pollution Control: The Case of Infant Health.” We presented some evidence that pollution levels in general and sulfur dioxide in particular had positive effects on race-specific neonatal mortality rates in a sample of large US counties in 1977.
In 2003 Ken Chay and Michael Greenstone published a paper in the Quarterly Journal of Economics entitled “The Impact of Air Pollution on Infant Mortality: Evidence from Geographic Variation in Pollution Shocks Induced by a Recession.” They corrected everything that Ted, Fred, and I did wrong.
Seriously, Ken and Mike produced a fascinating study which capitalizes on a quasi-natural experiment: namely, the 1980-82 recession resulted in substantial variation across counties in reductions in air pollution caused by total suspended particulates (TSPs). At the same time there were few confounding changes in other determinants of infant mortality. Using data for approximately 1,000 counties during this period, they find that a 10 percent reduction in TSP results in a 3.5 percent reduction in infant mortality. This implies that 2,500 fewer infants died in the period at issue than would have in the absence of TSP reductions. Their results suggest large benefits from reductions in pollution, both above and below the standard established by the Clean Air Act of 1970.
I cannot enumerate in 5 or 10 minutes all the reasons that I like this paper so just let me emphasize a few. Most of these deal with the methodology that Ken and Mike employ and with the detailed specification checks they perform to underscore the robustness and plausibility of their findings.
- They construct a time series of county cross sections for the period 1978-1984. This allows them to demonstrate why cross-sectional estimates are misleading because one cannot control for unmeasured county-specific effects. It also allows them to show that the association between changes in TSP and infant mortality rates is strong during the 1980-82 recession and nonexistent in the period before and after the recession. This is important because dramatic improvements in air quality occurred during the recession but not before or after it.
- They emphasize results from a first-difference instrumental variables specification which includes year and county fixed effects. They treat changes in air pollution and per capita income as endogenous and employ first and second lagged levels of these variables as instruments. This allows for the possibility that past shocks to infant mortality have feedback effects on current values of pollution and income.
- They show that pollution has no impact on external causes of death (accidents and homicides). This establishes the internal consistency of their findings since their is no obvious causal pathway from pollution to these causes of death.
- They also establish the plausibility of their findings by showing that the effects are much bigger for deaths within 24 hours and within 28 days of birth (the neonatal period) than for deaths between the second and twelfth months of birth (the post-neonatal period).
- To reduce biases due to confounding economic shocks, they show that there is no interaction between income changes and TSP changes in assessing the overall effect of pollution on mortality. In addition they compare changes in mortality and pollution among counties with little activity in manufacturing that do and do not border counties with high levels of manufacturing employment. The effects here are similar to those in the full sample.
All in all this is a most remarkable paper. It gives me great pleasure to present the Twelfth Annual Kenneth J. Arrow Award to Ken Chay and Michael Greenstone for the best paper in health economics published in 2003.
Trackback Pings
TrackBack URL for this entry:
http://www.healtheconomics.org/a/mt-tb.cgi/22