Mass appraisal: a statistical approach to determining the impact of a property’s attributes on its value

Abstract

One of the problems occurring when estimating the value of a property is determining how the individual attributes of a property affect its value. Until now, this was estimated either on the basis of analyses of prices and market features of similar properties, or by analogy to other properties in terms of the type and area of markets, by examining or observing the preferences of potential buyers, or in other ways (Polska Federacja Stowarzyszeń Rzeczoznawców Majątkowych, 2008). This article refers to the last option and presents an alternative way of dealing with the aforementioned issue. An approach based on statistical methods, using dependency coefficients is introduced as an objective proposal to determine the impact of individual attributes on the value of real estate. The idea of including statistical methods (dependency coefficients) in the procedure of mass real estate valuation is the basis of the statistical approach of the Szczecin Mass Real Estate Valuation Algorithm (Pol. Szczeciński Algorytm Masowej Wyceny Nieruchomości – SAMWN). Considering the specificity of the features (attributes) describing real estate, which in most cases may be presented on an ordinal scale, four coefficients measuring the strength of the relationship between the features and the value of the property were proposed. Since previous studies, e.g. Dmytrów et al. (2020), indicated that the results obtained for partial coefficients reflect the relationships found on the market more efficiently, the analysis was extended to include partial variants of selected coefficients. Then, six procedures for calculating weights in SAMWN were proposed. The obtained results were subsequently ordered on the basis of linear ordering procedures, as described in the work by Strahl (1978) and Pawlukowicz (2010). The aim of the article was twofold. Firstly, to identify the coefficients of dependencies that can be used in property valuation based on SAMWN, and also in the analysis of those properties whose attributes are often described on an ordinal scale. The second aim was to identify a combination of methods describing the influence and weights of the attributes which under SAMWN procedure produces results closest to the property values determined by real estate appraisers.