Determinants of farmland values in the Campine region (Belgium): Conditional and unconditional quantile regression estimates

Abstract

In this paper, our aim was to go beyond conventional OLS estimation, and use quantile regression (QR) to estimate major determinants of farmland values in the Campine region, situated in the north of Belgium. The QR framework provides a pragmatic approach to examine the impacts of transaction characteristics along the entire price distribution. Our specific contributions were (i) the focus on soil pollution, and (ii) the use of (non-standard) unconditional QR (UQR) proposed by Firpo et al. (2009). The UQR approach is better suited to address policy issues and welfare implications of, for example, bioremediation (i.e., the use of any organism metabolism to remove pollutants). In this paper, we were primarily interested in estimating the hedonic (or implicit) prices of Cd contaminants present in farmland. It was found that Cd contamination has different effects at different points of the farmland-price distribution, with a significantly negative impact on the median price (or the central region of the price distribution). Unfortunately, though, due to the small size of our sample (e.g., only about 6.8% of the 511 observations used in the estimations have Cd level exceeding 2 ppm), the QR estimates were not always statistically significant and/or economically meaningful. Therefore, we plan to demonstrate the usefulness of both CQR and UQR in future work by using more extensive datasets.