Housing attributes and Hong Kong real estate prices: a quantile regression analysis

By nature, people's tastes and preferences are unique and diverse so that a constant coefficient of each housing attribute produced by ordinary least squares (OLS) is not able to fully describe the behaviour of homebuyers of different classes. To complement the least squares, quantile regression is used to identify how real estate prices respond differently to a change in one unit of housing attribute at different quantiles. Theoretically, quantile regression can be utilized to estimate the implicit price for each housing attribute across the distribution of real estate prices, allowing specific percentiles of prices to be more influenced by certain housing attributes when compared to other percentiles. Empirical results demonstrate that most housing attributes, such as apartment size, age and floor level, command different prices at different quantiles. With the use of this approach, the efficiency of the mortgage markets is enhanced by offering more accurate prediction of real estate prices at the lower and upper price distribution.