摘要
因为区域间经济收敛、外商直接投资和知识溢出等领域的空间经济计量研究依赖于空间关系的存在,所以进行空间相关性Moran’s I检验是关键。然而,已有空间相关性Moran’s I检验理论受到众多假设条件限制。利用"名义水平—实际水平"图和"名义水平—功效"图,解析非对称Wild Bootstrap方法用于空间相关性Moran’s I检验的有限样本性质,发现即使模型不满足经典的分布假设条件,与渐近检验相比,Bootstrap方法也能够有效地检验研究对象间的空间相关性。
Spatial econometric analysis of regional economic convergence,FDI and knowledge spillover relies on spatial dependence among regions.Its key is Moran's I test for spatial dependence.However,Moran's I test theory is restricted by many assumed conditions.In this paper,we apply the residual-based asymmetric wild bootstrap method for testing spatial dependence in a linear regression model.Finite properties of the bootstrap Moran test are investigated using the "nominal-size vs.actual-size" and "nominal-size vs.power" plots.These plots show that the bootstrap test for spatial dependence among research objects is effective relative to the asymptotic approach when the classical distributional assumption is not warranted.
出处
《统计与信息论坛》
CSSCI
2011年第2期3-8,共6页
Journal of Statistics and Information
基金
国家自然科学基金项目<空间经济计量模型中Bootstrap方法有效性研究>(70871041)