• •

### 基于半参数混频误差修正模型的中国CPI预测研究

• 出版日期:2018-10-25 发布日期:2018-10-22

### Research on China CPI forecast Based on Semi-parametric ECM-MIDAS Model

Lu Wanbo & Yang Dong

• Online:2018-10-25 Published:2018-10-22

Abstract: Considering that the macroeconomic variables have obvious nonlinear characteristics, a semi-parametric ECM-MIDAS model is constructed for the data which exist cointegration relation in the MIDAS model. By using the GLR test, the problem of the consistency test of the functional form of the parametric regression model is solved. The simulation result shows that SEMI-ECM-MIDAS has better forecast performance when the error term is nonlinear. In application part, China's stock market weekly data, China’s monetary market and crude oil market international data are used to make short-term forecast of China's CPI. A comprehensive comparison was made among the models. The results show that the error correction term has obvious nonlinear characteristics. The semi-parametric ECM-MIDAS model proposed in this paper always has the best prediction accuracy. In addition, the prediction results are not affected by the choice of dynamic mixed frequency cointegration relationship.