The Seasonal Adjustment Of Economic Time Series

Seasonal factors are obviously included in the sub-annual economic time series. Those seasonal fluctuation are caused by the changing climate, religious practices, social customs and business practices will influence the true tendency of time series and mislead the decisions-makers. One of the main method to remove incomparability of seasonal factors is using index over same period of last year which is a method proved to have limitations. International research on the seasonally adjustment is very deep, many national statistical agencies and banking institutions led the research work contains the research method of seasonal adjustment and develop and upgrade the seasonal adjustment software. Up to now, it has formed two method systems which are popular in many countries. Many countries have attached great attention to the practice of seasonally adjustment such as periodically publish data be adjusted.In order to accurately understand economic trends, we should adopt international standards of data publishing. At the same time, those methods of adjustment are developed on local seasonal features, we should make research on seasonal adjustment on our own features.At first, the paper analyzes the popular seasonal adjustment methods x-12-ARIMA and TRAMO/SEATS, and then display the feature of our own country. Based on the method systems and seasonal features, adjust the series of total retail sales of social commodities. Seasonal adjustment method is divided into three modules. Three modules are adjusting the calendar holidays, the traditional methods of seasonal adjustment system containing compares of different method, seasonally adjustment considering the Spring Festival effects. Each stage can be carried out comparative method. Finally we can get the best way to adjust the seasonal factors.Our vacation length have changed, this paper proposed an innovative idea that can add multiple dummy variables in the model.Finally, the article to the conclusion (1)It is better to add multiple dummy variable model, the article got the holiday effect size.(2) In x-12-arima system,the model in consideration of the trading day is the best multiplicative model. And T/S is better than x-12-arima. (3) Spring Festival effects contain the effect before the Spring Festival and the effect during the Spring Festival but not the effect after the Spring Festival.