Construction Price Level Forecasting In China

Construction price level forecasting, which is the indispensable components of the construction price management system, is critical to investment and development strategy of the government and stakeholders of the construction market. In practice, little research had been made for construction price forecasting, not only for the methods of static cost directly or static cost supplemented with inaccurate or outdated price adjustment index had been used to forecast or control cost, but also for the reliance of construction cost management personnel to the Quota and fee standard. It is obvious from the literature that methodologies employed to forecast domestic construction price mainly include those not heavily dependent on statistics, such as Grey System, Neural Network and Fuzzy Mathematics, which are all belong to "Black Box" forecasting system, and useful while statistics is in shortage, but it always fail in giving reasonable explanation to the result, sometimes, its confidence level is always in doubt. Therefore, theories such as Economics, construction economics, forecasting methodology were employed to forecast domestic construction price level with time series forecasting model, multi-regression model with independent variables and VAR(2) model, which take dynamic of the economic system into full consideration. Through all research work of this paper, following conclusion could be obtained:1. Forecasting model make full use of advantages of traditional time series and random time series techniques is established to improve forecasting accuracy of domestic construction price time series, namely, the trend of the time series is estimate firstly with exponential smoothing model characterized with its advantages of adjusting trend of the series continuously, then, random time series is employed to extract information from the residues of former estimation. At the same time, In order to forecasting regional construction price of China with the methodologies discussed in this paper, taking nature of domestic regional construction price of short time series span and rich cross-sectional information into consideration, panel data was employedto make full use of cross-sectional data and time series data to established panel data construction price forecasting models for all 29 demotic regions of China.2. Continuous and stable leading indicators of construction price were identified with modified Granger Causality with nature of domestic construction price taken into consideration and dummy variable representing marketing degree of the construction market used, shortcoming of choosing independent variables of multi-regression models with experience is overcome, it is found that there are two major structural change in domestic construction industry in year 1984 and year 1991, and after year 1991, marketing degree of domestic construction market have been comparatively perfect, positive result of market economy theory has been obvious in it. According to which, domestic construction price multi-regression model has been established, from which, it is found that large and medium-scale project investment is the most remarkable factors to influence construction price, the price level of construction materials and per capital GDP are the second and third prominent factors to influence construction price, the result is useful in explaining reason of construction price movement and forecasting of short-term domestic construction price.3. Based on Vector Auto-regression theory, VAR(2) model with variables of construction price, number of large & medium construction projects in progress, number of construction companies, unemployment ration of the town and city, amount of M2 supply and number of large & medium construction projects completed is established, it is found that domestic construction price has some viscidity in nature through impulse analysis and variance decomposition; through cointegration analysis, it is found that there is some cointegration existing among variables is the system, through Vector Error Correction...