It has been widely acknowledged that real estate prices are spatially autocorrelated and correcting for spatial autocorrelation has become a standard practice in hedonic pricing models to improve estimation efficiency. However, its causes have not yet been fully explored. This paper puts forward by hypothesizing that spatial autocorrelations in housing prices are determined by market volatility: low market volatility weakens spatial autocorrelations and vice versa. The hypothesis is tested using data for over 150 000 transactions of homes sold between 1997 and 2009 in Hong Kong. The empirical results confirm the volatility explanation for the spatial autocorrelations. The finding contributes to a better understanding of the causes of spatial autocorrelation.