统计研究 ›› 2012, Vol. 29 ›› Issue (6): 99-105.

• 论文 • 上一篇    下一篇

一种新的空间权重矩阵选择方法

任英华 游万海   

  • 出版日期:2012-06-15 发布日期:2012-06-20

A New Selection Method of Spatial Weight Matrix

Ren Yinghua & You Wanhai   

  • Online:2012-06-15 Published:2012-06-20

摘要: 空间权重矩阵选择问题一直是空间计量经济学中的一个难题,权重矩阵的选择正确与否关系到模型的最终估计结果。本文在空间滞后模型框架下,把空间权重矩阵选择问题转化为变量选择问题,然后利用CWB方法进行变量选择。中国城市服务业集聚机理实证研究显示,利用本文所提出的方法所选取的空间权重矩阵较为合理,进而可以减少因为空间权重矩阵误设问题而引起的模型估计偏误。在大样本情形下,该方法可以非常有效地降低计算成本。

关键词: 空间滞后模型, 空间权重矩阵, Component-wise Boosting方法

Abstract: Spatial weight matrix will affect the estimation results of the model. How to select an appropriate spatial weight matrix is an important issue in spatial econometrics. The paper applies a component-wise boosting algorithm to deal with the selection issue of a spatial weight matrix in spatial lag models. It shows the selection issue of spatial weight matrix can be reformulated as a variable selection problem and then use the component-wise boosting method to select variables. The empirical results from the mechanism of urban service industry aggregation demonstrate that the chosen spatial weight matrix is rational and using it can reduce the estimated bias effectively. Besides, the method can reduce the computation cost effectively in large sample cases.

Key words: Spatial Lag Model, Spatial Weight Matrix, Component-wise Boosting Algorithm