具有GARCH(near-IGRACH)-normal误差项时序的ADF单位根检验

Testing for a unit root in time series with GARCH(near-IGARCH)-normal errors

在有限样本情况下,随机模拟具有GARCH(near-IGRACH)-normal error的ADF单位根检验,分析了两种数据生成模型下不同显著水平对临界值的影响.结果表明:随着显著性水平值的下降,相应的临界值绝对值变大了,特别是使用Zρ检验统计量时,对应的size distortion越来越严重,对高频金融时序进行单位根检验时,选择Zρ或Zt统计量需要进行一个权衡.

Under the limited samples, the author simulated the ADF unit root test of finite - sample time series with GARCH（ near - IGARCH） - normal error. The effects of the critical values unit root test were analyzed under the sorts of significance level and two kinds of models of the data generating. It shows that the absolute critical values become bigger as the significance level values are lowing, especially when using the Zp to test statistic and calculation and its size distortion is more serious. So the anthor may think that the Zp or Zt statistics is selected as the unit root testing for high frequency financial time series.