混合地理加权回归模型中的空间相关性检验和参数估计研究  

A Study on Spatial Correlation Test and Parameter Estimation in Mixed Geographically Weighted Regression Models

作　　者：乔宁宁[1] 

机构地区：[1]厦门大学经济学院统计系

出　　处：《数量经济技术经济研究》2013年第8期93-108,共16页Journal of Quantitative & Technological Economics

基　　金：国家社科重大基金研究项目(08&ZD034);国家社科重点基金研究项目(09AZD045);教育部哲学社会科学研究重大课题攻关项目(11JZD019)资助

摘　　要：对混合地理加权回归模型，提出新的空间相关性检验统计量，利用三阶矩Х^2逼近方法导出其检验P值的近似计算公式，模拟结果显示该检验统计量在检测空间相关性方面具有满意的功效。为了处理数据中可能同时存在的空间相关性和空间异质性，引入一类新的混合地理加权空间滞后回归模型，模拟结果表明该估计方法具有较高的可靠性和稳健性。与全局空间滞后回归模型比较，混合地理加权空间滞后回归模型在处理空间异质性方面具有更优良的表现。Based on the mixed geographically weighted regression model, a new testing statistic for spatial correlation is constructed. The corresponding approximation of p-value is derived by employing the third-moment Х^2 method. The simulation results demonstrate that the proposed test has satisfactory power in detecting spatial correlation. In order to deal with the simultaneous existence of spatial correlation and spatial heterogeneity in a dataset, we introduce a new class of mixed geographically weighted spatial lag regression model. The simulation study shows that the estimation method has good reliability and robustness. Comparing with global spatial lag regression model, the new model has better performance in processing spatial heterogeneity.

关 键 词：MGWR模型 空间相关性 空间异质性 

分 类 号：F224.0[经济管理—国民经济]