A Predictive Study On The Inflation Uncertainty In China Based On MS-GARCH Model

As one of the most important economic variables,inflation is closely related to national development,social stability,and people's welfare.The volatility of inflation volatility or inflation uncertainty is as important as the level of inflation.Even if all prices in the economy have sufficient flexibility,it can cause severe welfare losses to risk-averse economic agents.Therefore,it is essential to be able to predict inflation uncertainty as accurately as possible.The prediction of inflation uncertainty usually depends on the typical fact that the high-frequency time series data shows clustering of volatility.In order to capture these features,the ARCH / GARCH model is commonly used in the literature for prediction research.Because the structural form of the conditional variance of this type of model is relatively inflexible,the possible structural changes of the conditional variance are ignored.In order to solve this problem,this paper applies the MS-GARCH model to fit the situation that the volatility of inflation rate has different volatility structures under different mechanisms,and gives the predictive study of inflation uncertainty in China.The specific work is:Firstly,we conduct an empirical analysis of China's inflation uncertainty.We select China's monthly CPI year-on-year growth rate from January 1990 to December 2019 as raw data,and establish AR(1)-GARCH-N,AR(1)-GARCH-t,MS-GARCH-N and MS-GARCH.The models use ML as the parameter estimation method,and give the predictive results of inflation uncertainty for the previous 12 periods.In the model evaluation part,seven statistical loss functions are established to test the predictive performance of the four competitive models.The research results show that the simple AR(1)-GARCH-N model is better for short-term prediction.For the medium and long-term forecast,the performance of the MS-GARCH-t model is superior.Secondly,the Granger causality test is carried out on the relationship between the two,and it is found that the causal relationship between the two has a certain relationship with the number of lag periods selected: in the short term,the two are Granger reasons for each other.In the medium and long term,inflation is Granger cause of inflation uncertainty.The final is to use the vector auto-regressive model(VAR)to explore the dynamic impact of China's inflation uncertainty on inflation and its own.The research results show that the effect is significant for both inflation and itself in the short term,but will gradually weaken over time until it disappears.