The Consumer Price Index Model Empirical Analysis And Prediction Based On The VAR Model

Consumer Price Index(CPI) is not only closely related to people’s life, but also can reflects a country or a region’s operation conditions.Over the years many scholar have made various studies from different angles. But because the economy of each country is different, so the conclusion is not always the same.China’s consumer price index has experienced a rapid rise in 2007 and 2008, so this attracted a group of scholars to use econometric models to forecast the price level for the next few years, and to put forward some corresponding preventive measures against inflation.And the next two years’moving trend of the CPI was just as predicted by these models.It shows that the CPI is of both forecast accuracy and importance.But since 2012 the CPI displays a falling trend and then gradually gets stabilized, so it becomes important to research how long the trend of stability will be, and whether there will be a new round of deflation.So this paper sets about from the various factors that affect the CPI,according to China’s monthly CPI data from January 2003 to September 2014,uses VAR model to study the relationship between CPI and its influencing factors such as material purchase index currencies and monetary growth rate (M2) of the same period,and predicts quantitatively the CPI value in the future through using the established VAR model.The results of the study found that the CPI is more sensitive to its own movement. Based on this, this article selects the ARMA model, through the CPI values of past, current value, and lag of random perturbation terms weighting to depict well the changing rule of the CPI in the future, and also forecasts the consumer price index for 15 months in the future.Finally, according to the empirical statements and comparisons made in the previous chapters, it concluded that the overall level of China’s price will be steadily up-going in the future, but there may be a slight deflation. Also the VAR model’s prediction is better than that of the ARMA model. At last, it points out a few shortages of this paper.