中国季度GDP的季节调整：结构时间序列方法

 内容提要：本文利用结构时间序列方法讨论了中国季度GDP的季节调整问题，从季节单位根、季节自相关、周期自相关等多个方面对不同季节模式的调整结果进行了比较。结论认为，随机虚拟变量形式和三角函数形式得到的调整结果非常相似；结构时间序列方法更好地捕捉到了时变季节特征，明显优于X-11和SEATS方法；非高斯稳健季节调整的结果表明，高斯结构时间序列方法具有较好的稳定性。

Seasonal Adjustment of China Quarterly GDP

The paper analyzes the seasonal adjustment of GDP in China.We compare the adjustment through seasonal unit root,seasonal autocorrelation,periodic autocorrelation etc.Several important conclusions are drawn;the stochastic dummy variables form and trigonometric form get the similar results;The structural time series model captures the time-varying seasonality more precisely than X11 and SEATS;non-Gaussian model reveals that the Gaussian structural time series method is robust for seasonal adjustment.