One of the basic problems connected with land management is the question of accurate forecasting and valorisation of the real estate value. The statistical tools applied for this purpose are not always efficient due to certain imperfections in the land market. The present paper constitutes a proposal for the modification of statistical analysis methods aimed at adjusting them to the conditions of poorly developed local markets. A linear model of multiple regression has been used most often in the theory and practice of econometric modelling of transaction prices on the land market. However, its application is often ineffective due, for example, to considerable differences in prices and problems with an objective evaluation of quality factors affecting the real estate value (eg location). The research results show that better effects may be achieved by using minimisation of square sum of relative variations in actual and forecasted prices instead of the square sum of differences between actual and forecasted prices. Viewed from the perspective of valorisation and forecasting of the value of real estate, an important problem is valuation of short-term fluctuations in prices. In the paper this problem was solved by analysing data for individual time intervals with the help of modelling principles. The possibility of estimate errors arising from the application of the models obtained is also indicated. The research was conducted on the basis of the data obtained from a local market of landed property located in the city of Olsztyn in Poland.

This item is no longer available.