Estimation of Housing Demand Using Non-Linear Regression Model

Abstract

Many studies have been conducted to estimate the housing demand; however, some important factors were distorted or missing in previous studies: income and housing cost. In this context, this study included income and cost variables with a non-linear form, which has an assumption that the increase in income has a marginal influence on the housing demand as income increases with a certain growth and that it happens similarly in the case of the cost variable. This study can be largely separated into two steps: previous research review and empirical analysis. Regression analysis was conducted with both linear and non-linear forms according to housing possession types, owner households and rental households. This could show the difference in the influence by possession types between two hypotheses: the income and cost have a constant effect on demand for living space, or they have a marginal effect on the demand. The estimation reveals that the linear model for income and cost variables has a rapid growth trend for the future while the non-linear model for those variables shows a gentle growth tendency during the same period. Also, there will be more demand for owner households than for rental households in the future. The sensitivity analysis showed that the prediction of the annual growth rate of income and cost is the most important to exactly estimate the housing space in the future. This study has some implications as follows: the non-linear model is more realistic than the linear model for estimating reasonable housing demand; therefore, it is very important to choose appropriate variables and methods in the estimation process. Flexible approaches to estimation parameters such as the growth rate of income and cost for the future are needed due to the variance in economic conditions resulting from exogenous circumstances.