摘要
采用空间计量经济学的方法,研究我国31个省市1997-2006年间对外贸易的空间关系.提出了一种新的经济空间权重矩阵——基于区域变量协动程度的协动空间权重矩阵(Co-Movementspatial weight matrix),并分别采用简单地理空间权重矩阵和协动空间权重矩阵对31个省市间的空间关系进行了检验,说明我国各省市间存在空间相关关系.基于这两种空间权重矩阵及31个省市的数据,分别构造得到影响各个省市的空间滞后项,将其引入到误差修正模型中,对31个省市的出口数据进行了预测,并以有代表性的三个省市:广东、北京、湖南说明模型预测效果.结果表明:在一些省市的预测中,引入空间关系十分必要,并且能够显著的提高模型的预测精度.
Using techniques of spatial econometrics,we study the spatial relationship for 31 Chinese provinces and cities from 1997 to 2006.Based on the regional variables＇ comovement,a new economic spatial weight matrix-comovement spatial weights matrix was proposed and put into practice.For the comovement spatial weights matrix and the geographical contiguity weights matrix,we test the spatial correlation among 31 provinces and cities,and it shows that the spatial correlation exists.Based on the two spatial weight matrixes and data of 31 regions,we establish the spatial lag variable and introduce it into the error correlation model,forecast the export of 31 provinces and cities and choose 3 representative regions to give a demonstration.The empirical application on prediction of 3 representative regions shows that introducing spatial relationship is required and it can improve forecasting accuracy effectively in some regions.
出处
《系统工程理论与实践》
EI
CSCD
北大核心
2009年第11期84-92,共9页
Systems Engineering-Theory & Practice
关键词
空间相关性
协动空间权重矩阵
空间-误差修正模型
预测
spatial correlation comovement spatial weights matrix spatial error correlation model prediction
