Distance in geographic and characteristics space for real estate pricing
A common approach to predicting the price of residential properties uses the hedonic price model and its spatial extensions. Within the hedonic approach, real estate prices are decomposed into internal characteristics of an apartment, apartment characteristics and external characteristics. To account for the unobserved quality of the surrounding environment, price models include spatial price correlation factors, where the distance is usually measured as the distance in geographic space. In determining the price, a seller focuses not only on the observed and unobserved factors of the apartment and its environment but also on the prices of similar marketed objects that can be selected both by geographic proximity and by characteristics similarity. The purpose of this study is to show the latter point empirically.
This study uses an ensemble clustering approach to measure objects' proximity and test whether the proximity of objects in the property characteristics space along with spatial correlation explain the significant variation in prices.
In this paper, the pricing behaviour of sellers in a reselling market in Perm, Russia is studied. This study shows that the price transmission mechanism includes both geographic and characteristics spaces.
After testing on market data, the proposed framework for the distance construct could be used to obtain higher predictive power for price predictive models and construction of automated valuation services.
This study tests the higher explanatory power of the model that includes both the distance measured in geographic and property characteristics spaces.