The Period Length Identification Of Spring Festival And The Improved Adjustment Methods In The Seasonally Adjustment

The economic time series of monthly or quarterly are highly susceptible to moving holiday factors,eventually lead to not comparable between observed values.At the same time,it can also distort the periodicity of the series.At present,the seasonal adjustment method almost all contains the adjustment of moving holiday factors.The Spring Festival,as China's most important and widespread traditional festival,will have a profound effect on people's production,consumption habits and national policy.So this paper takes the Spring Festival as an example to study all aspects of the moving holiday effects.This study first carry on a simple introduction on the seasonal adjustment method,and then gives the moving holiday adjustment models and related attributes,Then combed the major moving holiday adjustment methods and classified them into three categories:X-13ARIMA-SEATS moving holiday adjustment methods with the original sequence as the dependent variable;Using the separation of irregular factors(I)as the dependent variable of Malaysia SEAM program and the nonparametric regression moving holiday adjustment methods:monthly average log method and proportional factor method.And then after combing the current mainstream moving holiday adjustment methods and specific process at home and abroad,we put forward a new idea of determining the period length of the Spring Festvival.Namely,first,we campare various possible period length of moving holiday,and then by some criterion proposed in this paper to selected the best period length.Second,a stochastic simulation to obey the time series of the ARIMA model and finally combining the characteristics of economic time series and the stochastic simulation series,generating a economic time series data that is not easy to get in real life,but do not break the rationality and with the Spring Festvival effects.,to verify the reliability of our best period length of the Spring Festvival we get before.This study selects the data between January 1993 and December 1993 of the total retail sales of social consumer goods to adjust the Spring Festvival effects,including Spring Festvival effects of recognition,Spring Festvival effects of adjustment and the period length of the Spring Festvival,etc.After the best period length of economic time series is determined,this study then extends the existing Spring Festival model,and put forward five extended Spring Festival model:3-period index function weighting moving Spring Festival method,3-period quadratic function weighting moving Spring Festival method,3-period trigonometric function weighting moving Spring Festival method,3-period double weights moving Spring Festival method and 3-period multifactor segmentation moving Spring Festival method.Among them,the 3-period double weights moving Spring Festival method use days scale factor and the observed value scale factor weights jointly set up the Spring Festival factor.To a certain extent,this method solve the issue of lacking data.And on the basis of the best period length determined before,by using the three moving Spring Festival models to conduct the moving holiday adjustment.Through the results,we can see that,for the total retail sales of social consumer goods series,the adjustment effects of the three main extended Spring Festival model is remarkable and is better than that of 3-period Linear moving Spring Festival method.It means that,the Spring Festival method this study proposed is not only has theoretical value,but also has practical value.On the other hand,fot the 3-period Linear moving Spring Festival model,by using the SAS software,this study independently created a program,that can output the regression explanatory variables of the Spring Festival holiday.when input different period length,the program can automatically generate the return of the Spring Festival holiday explanatory variables,makes the research more efficient.