Estimation Of Hedonic House Pricing Model Through Non-Parametric Methods: Istanbul Real Estate Market

The housing market is consisted of sub-markets that give it segmented structure and it exhibits heterogeneous characteristic due to both immovable and durableness of houses and their different structural, location and neighborhood characteristics from each other. Accordingly, each sub-market has different pricing structure owing to different supply and demand conditions of each individual market. Therefore, price estimation process for houses are rather difficult. However, Hedonic Pricing Approach especially used in pricing of products in heterogeneous markets that enables calculation of marginal impacts of individual characteristics on price of products and facilitates house pricing process. During investigation of the relationship between real estate prices and their characteristics, spatial effects are required to be taken into consideration which are significant determinant in house prices. Spatial effects can be presented as spatial autocorrelation and spatial heterogeneity. Whereas spatial dependency among house prices, which arise as a result of adjacency effect are described as spatial autocorrelation, variation in relationship between house prices and their characteristics based on the segmented structure of the housing market are described as spatial heterogeneity. Therefore during investigation of the relationship between house prices and their characteristics, there is need for models which allow relationships to vary according to locations and which consider spatial dependency in house prices. While Non-Parametric Spatial Models are appropriate for the standing need, the OLS Regression Model or Parametric Spatial Models do not take spatial effects into consideration because of their limiting assumptions. The present paper aims to have better understanding on demand side of the Istanbul Real Estate Market by investigating marginal effects of various characteristics of 2838 houses from 39 counties, which are assumed to represent Istanbul Real Estate Market through Non-Parametric Spatial Models based on Hedonic Pricing Approach in the period of October-December 2013. The results obtained from estimation through Non-parametric Spatial Models revealed that the relationship between house prices and their characteristics is differentiated for each sub-market and that there is non-linear relationship between house prices and spatial characteristics.