Abstract
The article discusses the topic of the application range of statistical methods in estimating property value on the grounds of a comparative approach. The analysis of application effects to estimate the unitary value of properties, respectively similar and dissimilar sets of market properties, by using the method of least squares and a linear price model. The prepared test set was developed from a priori assumed explanatory variable values as well as deterministically specified dependent variables (simulated prices) which were subjected to additional modification by a random factor. On the basis of the prepared set and series of accounting experiments, the estimation effects of any property out of a tested set were analyzed, understood as the determination of the value of the function of explanatory variables in the way of extrapolation or interpolation of values describing these variables. The experiments carried out show that the estimation of an explanatory variable for a random property out of a set of elements serving as the estimation base can be reliable only when it is related to the interpolation in the set of explanatory variables of this base. The application as an estimation base – a set in relation to which explanatory variables of the estimated property exceed the limits of corresponding variables, requires the completion of a basic set with records describing properties similar or close to the estimated property so that the values of explanatory variables for the estimated property are contained in the appropriate subsections of values of corresponding explanatory variables of the basic set.
The paper refers to the issue of defining property market value indicating, by the prism of conducted experiments, that the estimation results obtained by means of statistical methods do not always meet the requirements of the statutory definition of market value, and hear rather in the direction of a result corresponding to the so-called “desk appraisal” result.