Moran's I test of spatial panel data model — Based on bootstrap method
Moran's I test of spatial panel data model — Based on bootstrap method
复制标题
DOI：
10.1016/j.econmod.2014.04.022
复制
发表时间：
2014-08
影响因子：
4.7
通讯作者：
Tongxian Ren;Zhihe Long;Rengui Zhang;Rengui Zhang;Qingqing Chen
中科院分区：
文献类型：
--
作者：
Tongxian Ren;Zhihe Long;Rengui Zhang;Rengui Zhang;Qingqing Chen
在样本有限或分布误差项未知的情况下，面板数据模型的空间相关性检验一直是空间计量经济学尚未解决的问题。针对空间面板数据模型的Moran's I检验问题，采用快速双重Bootstrap（FDB）方法构造Bootstrap Moran's I检验，并通过MonteCarlo模拟实验，从尺寸失真和功效两个方面验证了方法的有效性.实验结果表明，在渐近Moran's I检验中，存在严重的尺寸失真，这种失真在Bootstrap Moran's I检验中是可以纠正的。
Under the condition of the finite sample or the unknown distributed error term, testing for spatial dependence in panel data models is an unresolved problem in spatial econometrics. In this paper, a fast double bootstrap (FDB) method is used to construct bootstrap Moran's I tests for Moran's I test in spatial panel data models, and Monte Carlo simulation experiments are used to prove the effectiveness from two aspects including size distortion and power. The experiment results show that, in asymptotic Moran's I test, there is serious size distortion, which could be rectified in bootstrap Moran's I test.