Parametric and non-parametric analysis of tax changes

In this paper, we examine the net effect of several major tax changes in Australia on residential property prices. Specifically, we consider the announcement and introduction effects that resulted from several policy changes including the introduction of the Goods and Services Tax (GST) and the accompanying First Home Owner Grant (FHOG). Using a large dataset of residential property sales in Melbourne, Australia, between 1992 and 2002 we estimate various models using parametric and non-parametric methods. While our parametric models suggest that the tax policy changes appear to have a statistically significant impact on house prices, no economically significant impact is detected by our non-parametric models, nor (upon closer inspection) by the parametric models themselves. Given the enormity of the sample size, this provides a telling example of the fundamental difference between statistical and economic significance and its implications for detecting government policy effectiveness