结构时间序列模型在季节调整方面的应用——与X-12季节调整方法的比较分析

建立一种基于结构时间序列模型的新的时间序列季节调整方法.首先,利用ARIMA模型研究时间序列的结构,根据序列的单整阶数(d)建立趋势循环分量的表达式,并在此基础上构建不同形式的结构时间序列模型.在结构时间序列模型中,针对经济指标分解得到的趋势、循环、季节及不规则因素是不可观测的变量,不能利用回归分析方法求解模型,因此,采用状态空间形式来求解模型.最后,利用结构时间序列模型对我国国内生产总值(GDP)和社会消费品零售总额等宏观经济时间序列进行了季节调整,并与目前广泛使用的X-12季节调整方法进行对比分析,实证结果表明,基于结构时间序列模型的季节调整方法具有相对较强的稳定性.

The Application of the Structure Time Series Model on Seasonal Adjustment——Compared with X-12 Seasonal Adjustment Method

In the paper,we construct a new seasonal adjustment method of time series on the basis of the structural time series model.By researching the structure of economic time series using ARIMA model,we firstly establish the expression of trend-cycle component according to the order of integration(d),and set up different forms of structural time series models.In the structure time series model,the economic indicator is decomposed into trend,cycle,seasonal and irregular components,which are unobserved and thus can't be estimated by classical regression way.So we estimate the model in the form of state space.Further,we use the model to decompose China's economic time series,such as GDP,Total Retail Sales of Consumer Goods,etc.Moreover we compare our model's results with X-12 seasonal adjustment method's,and the empirical conclusions show that the structure time series model is more stable when it is used to decompose seasonal component.