Comparative Study and Prediction of Seasonal Adjustment Methods for Economic Time Series

Abstract

Economic time series generally contain seasonal influences. In order to accurately analyze the development and changes of economic phenomena, internationally widely used methods such as X-12-ARIMA and TRAMO/SEATS are applied for seasonal adjustment to remove seasonal effects, thereby providing a theoretical foundation for further modeling, analysis, and forecasting of economic time series. This study takes the total retail sales of consumer goods in China from January 2009 to June 2023 as the research object. By comparing the residuals of two fitted models, it is found that the X-12-ARIMA method performs better. Therefore, the original series is modeled using the ARIMA approach, and data from July 2023 to June 2024 are forecasted, with prediction errors satisfying the national statistical error standards. After seasonal adjustment, seasonal factors are successfully separated from the total retail sales of consumer goods, restoring the underlying structure of the original series. This provides a basis for better understanding macroeconomic characteristics, capturing economic dynamics, and improving the accuracy of economic forecasting.