Seasonal Adjustment Methods Comparison Research

Seasonal adjustment is an important method to eliminate seasonal influence from the original economic time series in order to separate the trend, cycle, seasonal and irregular components. Seasonally adjusted sequence can be used to reflect the running situation of the national economy, do economic research and make macroeconomic policy. At present seasonal adjustment method research in China is lack of systemic mechanism, and has not yet developed a special method which is suitable for the characteristics of data structure in our country. Therefore, seasonal adjustment methods comparison research is of great significance in our country.Firstly, this paper induced the theoretical framework of X- 12-ARIMA,TRAMO/SEATS and structural time series model, then compared the advantages and disadvantages of them, and come to the conclusion: X-12-ARIMA method is simple in principle and has comprehensive functions and applications. For time series with standard seasonal characteristics, it has a good applicability, but the defect is the lack of flexibility in the treatment of some special seasonal problems for the nonparametric method. TRAMO/SEATS program’s advantage is the flexibility of setting regression variables, the simplicity of operation, a variety of inspection functions. For series with huge outliers and large amount it has strong adaptability. The advantage of structural time series model is to define the components with more flexibility, more intuitive model decomposition, and it also provides a good way of thinking to deal with seasonal heteroscedastic problem. The disadvantage is that model building is theoretical and subjective which is difficult for understanding and application, and the operation is time-consuming. Secondly, in the aspect of application, this article used three methods to make seasonal adjustment for the CPI sequence and compared the results, and come to the conclusion: the trend of structural time series model separated component is smoother and result is more stable. After measuring the CPI sequence link index by using three seasonal adjustment methods, found that X-12-ARIMA and TRAMO/SEATS seasonal adjustment methods are more sensitive to economic volatility and suitable for studying the economic turning points, seasonal adjustment method based on structural time series model is suitable for studying the long-term economic trends.