Abstract:
The traditional multiple regression analysis is widely acknowledged as a reliable technique and acceptable forecasting performance. This study examines the role of population growth, household income, and transaction volume in determining residential housing prices. A causal model is developed for Hong Kong using quarterly aggregated economic variables for the period of 1980:1 to 2000:3. In order to assess the adequacy of the multiple regression analysis to housing prices forecasting, the model is evaluated on its predictive accuracy on out-of-sample forecasts for the period of 2000:4 to 2002:4. It is found that the housing prices behavior of the current period is affected by the events of previous periods. The results also show that the causal model has a good predictive power and can explain the causal relationships between variables.
