Comparative Analysis Of Seasonal Adjustment Based On CPI Series

Most time series,especially economic time series,are influenced by a lot of different factors.As it is the most direct evidence for how economy has been working,whether we can judge the economic condition is based on if we can figure out the main factor influencing economy or not.we usually decomposes the time series into four factors by seasonal adjustment with mathematical statistics,which gives us a more direct view of the basic trend of time series.Now,seasonal adjustment methods are usually divided into two groups by principle.The one is method based on models,another is method based on filter.Time series must be random if method based on models is taken,of which most famous is TRAMO/SEATS developed by Spanish bank.We can know from the name that the method based on filter is to decompose time series by different kinds of filters,where the key factor is to reduce spectral power band of the input.And the most famous is X-12-ARIMA method.But the most important is that both of them are developed by foreign research authority,which makes some moving holiday only in China are not considered.In result,we can't expect more from seasonal adjustment.In this paper,we take X-13ARIMA-SEATS as basic seasonal adjustment method,which is the newest achievement in seasonal adjustment compared with X-12-ARIMA and TRAMO/SEATS,for the advantages it has.Then we put factor represented spring festival influence into regression model,and take it out from the final result of seasonal adjustment of CPI series,which makes them comparative.