Forecasting Based On Theory Of Cointegration And Error Correction Model

This paper forecasts Runoff and Grain's production of China based on the theory of cointegration and error correction model.The essay has mainly studied about three aspects. In the first ,with unit root test of non-stationary time series and the error correction model, we have established the error correction model to the multivariable second-order non- stationary time series.In the second we present a method of the Second Songhua River's runoff on the upper and lower reaches based on unit root test, cointegration test, error correction model. The method of cointegration analysis is applied in year runoff data of the Baishan and Fengman gauging stations, and the double variable error correction model is set up, then we predict the year runoff data of Fengman ganging station between 1989 and 1998.The results suggest that the model based on cointegration and error correction model is suitable in forecasting.At last we analyze the data of Chinese grain's production, the consumption of chemical fertilizers, the planting area of grain from 1983 to 2000 and found that they are second-order non- stationary time series. At the same time The multivariable cointegration theory is applied in the establishment of error correction model about above three variables. The model is tested by one step forecasting data of 2001, the effect is good. At last it predict Chinese grain's production between 2002 and 2004. It suggests that the forecasting result is almost same with the real production.