In the late 1970’s, events at Love Canal and the Valley of Drums raised public concern over the health and environmental risks associated with contaminated waste sites. In response to these incidents, the US Congress established the Superfund program to clean up hazardous waste sites. However, the economic benefits from the Superfund program have been questioned. Previous tract-level analysis of hazardous waste remediation under the Superfund program has failed to detect evidence of economic benefits being capitalized into nearby median housing values (Greenstone and Gallagher, 2008). 
Our study exploits high-resolution restricted-access census block data to provide direct evidence that these benefits are, in fact, large but highly localized in space. Comparing blocks located near National Priority List (NPL) sites that received cleanup treatment with those near similar NPL sites that that were not cleaned, we find that median housing values appreciate by 19.0% for blocks lying < 1km and by 5.8% for blocks lying < 3km from those sites.
Recognizing that most analysts only have access to publicly available tract-level data, we also show that evidence of these localized effects can be found by examining the entire within-tract housing value distribution, rather than simply focusing on the median. Our tract level analysis detects larger appreciation at the 10th percentile of the housing values (18.2%) than at the median (15.4%), and the effect of cleanup becomes insignificant beyond the 60th percentile. Finally, our block and tract level results are confirmed by additional evidence from geocoded proprietary housing transactions data describing four large housing markets.  These data show explicitly that it is the cheaper houses within a tract are more likely to be exposed to waste sites within one kilometer, explaining their greater appreciation from site cleanup.
Our method makes it practical to apply well-established hedonic techniques to value public goods with highly localized effects using unrestricted tract-level data. For at least two reasons, this approach has the potential for wide application. First, a growing number of studies document that amenities or disamenities.Second, in practice, many hedonic studies are forced to rely on tract-level data because of its nationwide coverage and public availability (Hanna, 2007; Greenstone and Gallagher, 2008; Grainger, 2010).  Finer resolution data alternatives are often inaccessible – house transaction data are proprietary and expensive, or limited in coverage, while block-level data have restricted access.

***
This seminar has been jointly organized by FEEM and IEFE, Bocconi University.