当误差项不满足经典的正态独立同分布假设条件时,利用Moran’sⅠ统计量的渐近分布进行空间相关性检验的功效较弱.文中把Bootstrap方法用于空间经济计量模型的空间相关性Moran’sⅠ检验统计量,Monte Carlo模拟实验结果表明：从功效角度看,当误差项服从正态独立同分布时,Bootstrap检验与渐近检验同样有效,甚至优于渐近检验;当误差项不服从正态独立同分布且存在异方差时,Bootstrap检验能够有效地提高渐近检验的功效;当样本量较小时,空间相关系数和空间衔接结构等对功效有显著影响,尤其是在空间衔接密度较高的Queen矩阵和空间相关系数小于0的情况下,Bootstrap检验的功效显著大于渐近检验;当空间权重矩阵为Queen矩阵时,Bootstrap检验的功效曲线随样本量增大而从＂√＂型变成＂V＂型,对称性增强,空间衔接结构对Bootstrap检验功效的影响减弱.

When the error does not satisfy the assumption of normal independent distribution,the spatial correlation test based on the asymptotic distribution of Moran＇s I test statistic suffers weak power.In order to solve this pro-blem,Bootstrap methods are applied to the Moran＇s Ⅰ statistic for spatial correlation test in the spatial econometric model.The results of Monte Carlo simulation indicate that the Bootstrap test is as effective as（even better than） the asymptotic test when the error satisfies the i.i.d.normal distribution;that the Bootstrap test effectively improves the power of asymptotic test when the error does not satisfy the i.i.d.normal distribution and is heteroscedastic;that both the spatial correlation coefficient and the spatial contiguity structure significantly affect the power of Bootstrap test with small sample scale,especially,under the Queen matrix with dense spatial contiguity and the spatial correlation coefficient less than 0,the power of the Bootstrap test is obviously higher than that of the asymptotic test;and that,when the spatial weight matrix is the Queen matrix,the power curve of the Bootstrap test changes from ＂√＂type to ＂V＂ type,with the symmetry being increased and the influence of spatial contiguity structure on the power being weakened.