LAUSR

Abstract The vertical dimension of the residential apartment market has been largely overlooked in spatial hedonic studies. To adequately quantify the environmental impacts brought by various urban natural elements on apartment prices using spatial hedonic models, the selection of an appropriate weights matrix, which could adequately approximate complex spatial relationship amongst apartment units in high-rise housing markets, is an important methodological challenge. Using a micro high-rise neighbourhood in Guangzhou (China) as a case, we compare the model performance of a set of 3-D spatial weights matrices in capturing the environmental impacts of a polluted river running by the residential neighbourhood on apartment transaction prices. Three 3-D spatial weights schemes are examined, including distance-based, contiguity-based, and a hybrid distance-contiguity weights matrix. Our Monte Carlo simulations and empirical comparison consistently confirm that contiguity-based weights matrices outperform distance-based and hybrid weights alternatives. Adopting the appropriate second-order queen’s cube contiguity matrix in a 3-D Spatial Autoregressive Combined (SAC) model, the spatial heterogeneity of urban river’s environmental impacts along both horizontal and vertical dimensions is revealed. We find significant negative impacts associated with the proximity and visibility of the polluted river, which are strengthened for apartments located on the 10th floor and below. Our findings highlight the urgency of restoring polluted urban rivers and provide evidence-based information for local authorities and private housing developers to restore polluted rivers in order to improve environmental amenities.