Hedonic House Prices and Spatial Quantile Regression

Abstract

Despite its long history, hedonic pricing for housing valuation remains an active research area, and applications of new estimation methods continually push research frontiers. However, housing studies regarding Chinese cities are limited due to the short history of China’s free housing market. Such studies may, nonetheless, provide new insights given the nation’s current transitional stage of economic development. Therefore, in this research, we utilize publicly accessible sources to construct a new dataset for an emerging Chinese city – Changsha, and incorporate quantile regression and spatial econometric modeling to examine how implicit prices of housing characteristics may vary across the conditional distribution of house prices. Substantial variation exists across quantiles, suggesting that ordinary regression is insufficient on its own. Quantile estimates of a spatial-lag model show considerable spatial dependence in the upper and lower parts of the distribution but little dependence in the medium range. Several other interesting patterns are also found, and their intuitions and implications are discussed.