Semiparametric measurement of environmental effects

Abstract

This paper gives the results of a semiparametric analysis of pollution effects on housing prices using the Boston Housing Data. The exposition introduces the basic ideas of modeling pollution impacts with hedonic price methods, discusses the standard log-linear model, and then introduces nonparametric estimation and semiparametric index models. We focus on the intuitive content and substantive results of the semiparametric analysis. We find that the impact of pollution is smaller than that previously estimated, and varies dramatically depending on the status level of the community. We give various interpretations of the findings, and contrast our methods with those used in previous analysis of the Boston Housing Data.