Sensitivity Analysis of Real Estate Price Oscillations Model

The purpose of this study is to understand the reaction of real estate prices to changes in different factors. In order to achieve this, a previously developed system dynamics model of real estate prices in Istanbul has been used (Özbaş and Özgün, 2005). The model has been developed and scenario analysis is carried out to some extent. However, a structured experimental analysis of the model under different conditions and the response of price to these changes are not investigated deeply. As one of the most important steps in system dynamics methodology is the output analysis that creates a basis for improving the structure, it is essential to determine the variables to which the model is sensitive, by using a more formal experimental design analysis. A complete sensitivity analysis can uncover the factors that should be controlled or monitored more carefully to obtain desired behavior in terms of price. In this study, the sensitivity of period, amplitude and mean of price oscillations to the changes in some selected variables of the model is analyzed. Two experimental sensitivity analysis designs, namely Fractional Factorial Design and Latin Hypercube Sampling, are used to measure the sensitivity of the model. Moreover, making such a research allows comparing different methods for parameter sensitivity analysis. The factors that turn out to be significant are not the same for two methods (Fractional Factorial Design and Latin Hypercube Sampling). As a final result of the study, the common factors and interactions found in the two techniques can be safely thought as the significant effects on the response.