Real-time Monitoring And Nowcasting Of CPI

Price fluctuations can not only reflect the extent of macroeconomic conditions,but also have an impact on the living standards and welfare of residents.Price stabilization has always been one of the four major objectives of China's macro-control.Price monitoring and forcasting is the basic work for the government to grasp the economic operation,understand the market changes,strengthen the scientific,effective and predictable macro-control and price supervision,and is an important work related to economic development and the national economy and people's livelihood.The consumer price index(CPI)is an important economic index reflecting price fluctuation.Its real-time monitoring and nowcasting can provide a more timely reference for the government's macroeconomic control.In recent years,with the economy entering a new normal,China's economic environment has undergone significant changes.At the same time,the price fluctuation represented by CPI has also appeared new characteristics,showing a new trend of "microwave".With the change of the consumption structure of residents and the increase of the uncertainty of the international situation,the factors that affect the price trend in the future become more complex,which puts forward higher requirements for price monitoring and forcasting,and also faces greater challenges.Using advanced measurement methods to improve the timeliness and accuracy of CPI monitoring and forcasting is undoubtedly of great academic and applied value.The National Statistical Bureau usually releases last month's CPI data on or around the 9th of the following month,which has certain time lag,and can not fully meet the needs of timely analysis of price situation and timely and appropriate adoption of regulatory measures.With the continuous improvement of China's statistical system,before the monthly release of CPI,the relevant departments will release some high-frequency data with strong correlation with CPI,and these information will help to improve the real-time monitoring of CPI.Therefore,in this paper,the mixing data model is constructed by using the relevant daily data,weekly data and monthly data,which makes the monitoring and forcasting frequency accurate to the weekly.In this paper,the "food+non-food" dichotomy framework is used to monitor and forcast the two sub items in real time.On the basis of the results of each sub-item,combined with the proportion of food and non-food in the CPI,the real-time monitoring and nowcasting results of the overall CPI are obtained.The real-time monitoring part of this paper firstly introduces the principle of mixed-frequency dynamic factor model,as well as the selection of various high-frequency indicators,and then extracts the weekly consistent index which comprehensively reflects the fluctuation of food CPI and non-food CPI.Secondly,combined with the proportion of food and non-food in CPI,weighted average of the two weekly consistent index is used to obtain the high frequency consistent index of the overall CPI,so as to achieve the purpose of real-time monitoring CPI at the weekly level.The results show that the CPI consistent index of weekly food,weekly non-food CPI consistent index and weighted average obtained by this method are consistent with the overall trend of real food CPI,non-food CPI and CPI,which can not only identify the subtle fluctuations in the month,but also accurately monitor the trend and development trend of CPI in real time.In the part of nowcasting,based on the weekly consistent index of food CPI and non-food CPI,nowcasting models of food CPI and non-food CPI were constructed respectively.After getting the nowcasting values of each sub-item,the weighted average is used to get the nowcasting values of CPI,and the nowcasting results are compared with those of the benchmark model.The results show that the nowcasting model with weekly common factor is superior to the benchmark model in both food and non-food categories.This shows that the nowcasting effect can be improved by using the information of high frequency index of weekly,and it has significant advantages in forcasting accuracy and trend fitting.In addition,from the perspective of nowcasting frequency,the model can not only real-time forcast CPI at the weekly level,but also update the nowcasting results with the release of new data.It not only realizes the real-time dynamic forcasting of CPI,but also greatly improves the timeliness of forcasting.There are three innovations in this paper:First,the mixed-frequency dynamic factor model of CPI real-time monitoring and nowcasting constructed in this paper solves the problem of the limitation of traditional methods for data co-frequency and the insufficient use of observable information to a certain extent.Second,this paper uses the "food+non-food" dichotomy real-time monitoring and nowcasting analysis framework to achieve the layered monitoring and forcasting of CPI to a certain extent.Third,based on the ADS model framework,this paper improves the quantitative relationship between different frequency data to make it more suitable for the CPI monitoring and forcasting in China,and to some extent,it makes up for the shortcomings of empirical methods in domestic research.