随机系数面板数据单位根联合LM检验的稳定性研究

发布时间：2018-05-03 21:14

本文选题：初始值 + 稳定性　；参考：《数量经济技术经济研究》2016年10期

【摘要】：本文从初始值的视角研究其对随机系数面板数据单位根联合LM检验稳定性的影响。推导当初始值不是依概率有界的随机变量而是渐近不可忽略的变量时联合LM检验统计量的渐近分布,并发现联合LM检验不再服从其原分布,且其与初始值有关,表明联合LM检验统计量的渐近性质不稳定。蒙特卡洛模拟结果显示,在有限样本情形下,初始值可能会导致联合LM检验统计量出现较严重的水平扭曲现象,即联合LM检验的统计性质不稳定,易受初始值影响。
[Abstract]:In this paper, the effect of initial value on the stability of unit root of random coefficient panel data combined with LM test is studied. The asymptotic distribution of the joint LM test statistic is derived when the initial value is not a random variable which is bounded by probability but is asymptotically not negligible. It is also found that the joint LM test is no longer subject to its original distribution and is related to the initial value. It is shown that the asymptotic property of the combined LM test statistics is unstable. Monte Carlo simulation results show that in the case of limited samples, the initial value may lead to serious horizontal distortion of the joint LM test statistics, that is, the statistical property of the joint LM test is unstable and vulnerable to the influence of initial values.
【作者单位】：西南财经大学博士后科研流动站;西南财经大学统计学院;
【基金】：中央高校基本科研业务费项目“时间序列前沿研究团队”(JBK150501);中央高校基本科研业务费项目“金融大数据研究基地”(JBK140403)的资助
【分类号】：F224

【相似文献】