Radosław Cellmer, Katarzyna Szczepankowska

Simulation modeling in a real estate market

View full text in PDF format

Abstract

One of the main features of the real estate market is its imperfection, to which belongs, among others, the lack of sufficient data on transactions, often preventing the use of quantitative methods to the analysis. Some solution of the problem of insufficient amount of data can be an expansion of a research area. However, it also means that, the analysis is made on a much diversified data set, so it is difficult to meet the condition of similarity between objects. In the face of the absence of sufficient number of market data, especially in less developed markets, an attempt was made to simulate a transaction that provided at one hand, a necessary data to build models of prices and value, and at the other hand, to evaluate various opportunities and options shaping the local real estate market. In the presented paper a theoretical basis for simulation modeling was done. Also such simulation was presented. It was based on the classical model of linear, multiple regression in an analytical form, and also with the use of Monte Caro simulation. In the study an assumption was made that, the probability of the next transaction will depend on the density of current transactions. In order to determine the density a nuclear estimation was used. It allows to explicitly taking into consideration a spatial resolution. As a result of the research, simulated transaction data, in a several input assumptions, were obtained, which fill the information gap on the researched fragment of the real estate market.

Keywords: simulation modeling; real estate market; Monte Carlo.