一种辨识投入产出模型中主系数的方法

目前在投入产出模型中四种常用辨识主系数的方法都是针对国民经济主系统而言,而随着我国经济的发展,一些具有相同经济特征的部门组成的国民经济子系统正在国民经济中的地位日趋显著,辨识国民经济子系统的主系数也就显得尤为重要.本文依据直接消耗系数的剔除将会使子系统蕴含受影响对象的信息量遭受损失,没有被剔除的系数集应蕴含受影响对象最少的信息量,以此达到剔除主系数损失最大的思想提出了损失量法,用以辨识国民经济子系统的主系数,并对2002年中国投入产出表的农业系统作了实证分析.同时将损失量法扩展到辨识国民经济主系统的主系数方面,并和信息量法作了比较,证明了损失量法能够比信息量法以更少的主系数达到相同的阈值,即在相同的阈值下损失量法选出的主系数个数比信息量法少.

Identifying Important Coefficients in Input-Output Model by Losing Information Method

This paper brings forward losing information method to identify important coefficients of national economic sub-system,which is different from major four methods that are all aimed at national economic main system.Its main idea is that the elimination of important coefficients will lose the most information of a system.We applied this method to agricultural system of 2002 Input-Output table and in contrast with containing information method,we find that,to reach the same limit,less important coefficients are sought out by losing information method than by containing information method.