空间经济计量模型Bootstrap检验的水平扭曲

发布时间：2018-05-25 19:35

本文选题：水平扭曲 + Bootstrap检验　；参考：《数量经济技术经济研究》2009年01期

【摘要】：本文使用回归残差的Bootstrap方法,对线性模型空间相关性进行检验。基于空间相关性检验统计量Moran’sI,在不同Bootstrap样本数及不同空间衔接结构下,研究并比较Bootstrap和渐近检验方法。通过MonteCarlo实验揭示了当空间经济计量模型中残差不满足经典正态假定条件时,空间相关性的渐近检验理论不再有效。本文把Bootstrap方法用于空间相关性检验,对水平扭曲进行了分析校正。研究同时发现,从水平扭曲角度来看,无论残差是否满足经典正态假定条件,空间经济计量模型Bootstrap检验通常都很有效。
[Abstract]:In this paper, the Bootstrap method of regression residuals is used to test the spatial correlation of linear models. Based on the spatial correlation test statistic Moran Si, the Bootstrap and asymptotic test methods are studied and compared under different Bootstrap sample numbers and different spatial cohesion structures. The MonteCarlo experiment shows that the asymptotic test theory of spatial correlation is no longer valid when the residual error in the spatial econometric model does not satisfy the classical normal assumption. In this paper, the Bootstrap method is applied to the spatial correlation test, and the horizontal distortion is analyzed and corrected. At the same time, it is found that the spatial econometric model Bootstrap test is usually effective from the angle of horizontal distortion, whether or not the residuals satisfy the classical normal assumptions.
【作者单位】：华南理工大学经济与贸易学院;美国波特兰州立大学;
【分类号】：F224

【相似文献】