Estimating the effect of air quality: Spatial versus traditional hedonic price models

Empirical studies of hedonic housing prices show that the spatial maximum likelihood estimation (MLE) method is preferable to the traditional ordinary least squares (OLS) hedonic method. Current computing capabilities restrict the MLE method to relatively small data sets. This paper circumvents this limitation by coupling the spatial MLE method with block bootstrapping, a form of Monte Carlo simulation that accounts for spatially dependent data. Blocks are created based on monthly and census tract information for resampling. For each month, we obtained 50 resamples of 750 observations from a data set of 15,727 residential properties to compare OLS and MLE empirical results. We find that the spatial MLE method consistently outperforms the traditional OLS method under these simulated conditions and that air quality matters irrespective of the method used.