구조모형과 시계열모형에 의한 부동산경기변동의 분석과 예측에 관한 연구 = (A)study on the analysis and forecast of real estate business cycle using structural model and time series model

국문 초록 (Abstract)

영문초록: The movements in real estates price are very complex and seem to be unpredictable. but many people want to forecast it. there have been many efforts to exploit models which could explain the movement of real estate price and could forecast it accurately. these models can be grouped in two categories: structural and time series models. The structural model have been good in explaining the fact but not always useful for forecasting real estate prices. and the time series models have been able to do a better job in forecasting real estates prices but not always useful for explaining it. In this paper I compared the advantage and disadvantage of models and I set up one structural model and two time series models. and this study aims to analyze and forecast the movements in real estate prices by models First, I set up new structural model using logistic difference equation and appraise the usefulness of the model by inputting real data in Korea to the model. and I could explain the past real estate business cycle and real estate policy in Korea more effectively by this model. Second, recently artificial neural network has received a special attention by financial researchers. but in the study of real estate business cycle, there has been no research for which it have been used. I set up time series model using artificial neural network. Third, I set up a Vector Auto-Regressive(VAR) model in order to analyze and forecast the housing market. the VAR model is simple compared to the structural models and exposes the relationships among other affecting variables better than the ARIMA models. It can be also used to analyze the response of a target variable resulting from the impulse of a specific variable. by impulse-response analysis I could study the relationships among interest rate, inflation rate, Chunse price, past house price and present house price in Korea. Next, I compare the performance of forecast for Kookmin Bank House Price Index using the VAR model and the ANN(artifitial neural network) model. the nonlinear artificial neural network model and linear VAR model are used to generate the out-of-sample competing forecasts for house price index. the result suggest that the out-of-sample artificial neural network model forecasts is more accurate than those which the VAR model forecasts. I hope that this study be helpful to the study of analysis and forecast of the movements in real estate prices using models