Simulation of the cost of residential property in Krasnoyarsk for 2013-2014

Data on the cost of secondary residential real estate in the city of Krasnoyarsk for the years 2013 and 2014 was collected and analysed. The number of objects in 2013-4031 items, in 2014-1037 items. Each flat is characterised by 13 parameters: number of rooms, layout, residential district, floor, total number of floors in the building, material of the walls, telephone, total useful floor area, living space, kitchen area, WC-and-bathroom unit, kitchen stove, availability of a balcony or stanza. On the basis of ordinary least squares (OLS) an adequate linear model of the cost of a flat for the years 2013 and 2014 was created. Their own significant factors were found for the models of each year. Also separate models for flats in Oktyabrsk and Sverdlovsk districts of the city of Krasnoyarsk for the year 2014 were created. It was examined how the influence of general factors on price formation changed. The number of rooms factor started to have a smaller effect on the price of residential real estate. The negative coefficient on this parameter increased by 1.5 times. Layout also started to have a smaller effect on the price. The negative coefficient increased by 2 times. Residential district: the negative coefficient on this parameter has become smaller which indicates that the location of the flat in the city has started to have a larger effect on the price. Total useful floor area: the effect of this factor did not change. Living space: a change in this factor’s coefficient turned out to be the most prominent because it increased by almost 5 times. From the analysis of models of the price of flats in Oktyabrks district one of important factors is availability of a balcony or stanza. For flats in Sverdlovsk district important factors are material of the walls of the building and availability of a telephone. These facts can be explained by the specific characteristics of these districts. When analysing residuals of the obtained linear models heteroscedasticity of the residuals can be seen; it was shown by Goldfeld-Kvant, White and Glazer tests. With the help of generalised least squares (GLS) on the basis of the linear model new models with homoscedastic residuals were created. The created models adequately describe experimental data.