季节异方差序列的季节调整问题研究

季节调整是从经济序列中剔除季节成分的重要方法。季节异方差的存在，使经典的季节调整方法无法彻底分离出季节成分，致使季节调整失败。本文针对季节异方差问题提出用于季节调整的改进的HS模型，并定义改进的HS模型构造季节异方差检验LR统计量，通过蒙特卡洛模拟方法分析该检验的检验尺度和检验功效。最后，利用我国税收总额月度序列给出实证分析，并通过对比考察了改进的HS模型方法季节调整的有效性。

Research on seasonal adjustment problem of seasonal heteroskedasticity

Seasonal adjustment is an important method to reject the seasonal factors from economic series. Classis seasonal adjustment methods become less useful when seasonal heteroscedasticity exists, which causes seasonal factor separated incompletely. This paper improves the HS model to solve the problem of seasonal heteroscedasticity, and then uses the improved model to construct a seasonal heteroscedasticity LR statistics and gets the size and power of the test by Monte Carlo method. At last, it examines the effect of the improved HS model by using it in China’s accomplished tax series seasonal adjustment.