Simulative Verification of the Possibility of using Multiple Regression Models for Real Estate Appraisal

Abstract The possibility of using multiple regression models in real estate valuation is the subject of disputes, both in theory and in practice. Econometric modelling is a difficult process, since a number of issues of substantive and numerical nature occur during that process. Modern technologies enable quick and easy model estimation with the use of virtually any quality of data. Naturally, it provokes property appraisers to use such models in the practice of real property valuation, particularly in mass appraisal, frequently without taking those issues into account. Consequently, the models obtained and applied in practice turn out to be of poor quality and, objectively speaking, should not serve as the basis for determining real estate value. The specificity of the real estate market and of the real properties themselves as objects traded in that market additionally exert a negative impact on the quality of the obtained models. In this article, the authors present the results of research which involved a simulation of various types of disturbances of a model artificially developed database of real estate prices and attributes as well as their impact on the quality of estimated models. The research will make it possible to answer the question of the degree and type of disturbances that are permissible in the functioning of a real estate market if the estimated models are to still satisfy the qualitative requirements defined for them, and thereby produce accurate valuation results. A model database will be disturbed by the deviation of prices from model prices and by reducing its size. Radom generators were used to obtain database disturbances.