平稳自回归模型的经验似然检验

Empirical Likelihood Test for Stationary Short Memory Time Series Models

这篇文章中,我们对平稳短记忆时间序列模型中的参数假设检验运用经验似然方法,在实际中,我们可能不仅关心所有参数的显著性而且更关心模型中的某一部分参数是否存在,因而我们除了建立检验所有参数的统计量,还建立了检验部分参数显著与否的统计量,这些统计量被证实都渐近服从卡方分布,另外,我们的模拟研究了检验的势函数并证实了提出的检验程序的有效性.

In this paper,we propose empirical likelihood method for parameter hypothesis test in short memory time series models.In practice,we may pay attention to not only the significance of all the parameters,but also the significance of some parameter in the models.So we construct different test statistics in these two situations,which are both shown to follow chi-square distributions asymptotically.In addition,our simulations investigate the power function for testing the concerned parameters and verify the validity of the proposed testing procedure.