一种新的空间权重矩阵构造及模拟比较研究

空间权重矩阵的构造一直是地理学、空间计量经济学和医学等领域中的难点和热点问题.权重矩阵合适与否,将直接关系到模型的最终估计结果.基于AMOEBA过程,通过局部Moran'I的比较,得到新的"邻接"关系区域,并结合基于经济距离的权重,构造出新的权重矩阵.为了进一步分析其优劣,利用蒙特卡罗模拟,在空间滞后模型下,对几种常用的权重矩阵,通过极大似然估计,利用评价标准进行比较研究.模拟结果表明:在大样本情形下,构造的新的权重矩阵,较其他几种常见权重矩阵对应的估计值更接近真值,模型拟合效果更好.同时,通过模拟比较研究也得到了一些有意义的结论.

A New Spatial Weight Matrix Structure and Simulation Comparison Research

The construction of spatial weight matrix has been a difficult and hot issue in the fields of geography,spatial econometrics and medicine.Weight matrix is appropriate or not,will be directly related to the final results of the model.In this paper,we based on the AMOEBA process,through the comparison of the local Moran'I,get the new "adjacency"relationship area,and combine the weight based on the economic distance to construct a new weight matrix.In order to further analyze its advantages and disadvantages,this paper uses the Monte Carlo simulation,in the space lag model,several commonly used weight matrix,through the maximum likelihood estimation,the use of evaluation criteria for comparative study.Simulation results show that:In the large sample case and in this paper,the structure of the new weight matrix,compared with several other common weight matrix corresponding to the estimated values of the close to the true value,the model fitting effect is better.At the same time,some meaningful conclusions are obtained through simulation and comparison.