摘要
GRAS方法目前仍是在投入产出表或社会核算矩阵的更新方面使用较为广泛的方法,但其仍具有改进空间。为此,文章构建了TGRAS方法,从投入产出表的更新结果出发,分析并检验了GRAS方法和TGRAS方法的更新精度。研究结果表明:(1)运用GRAS方法更新后,总产出(或总投入)越小的部门,更新结果的绝对百分比误差越大,即存在不对称性。(2)通过增加总产出(或总投入)较小部门的信息,TGRAS方法的更新结果在信息丢失指数、信息相关指数、标准化总百分比误差和离差距离指数方面均小于GRAS方法,在一定程度上降低了GRAS方法更新结果的不对称性,使得TGRAS方法的更新精度明显优于GRAS方法。(3)通过长短期记忆网络模型增加总产出(或总投入)较小部门的信息,TGRAS方法仅在信息相关指数上有所上升,TGRAS方法的更新精度仍然优于GRAS方法。
The GRAS method remains one of the most widely used approaches for updating input-output tables or social ac-counting matrices,but there is still room for improvement.In this context,this paper constructs the TGRAS method,and analyzes and tests the update accuracy of GRAS method and TGRAS method based on the update results of input-output table.The results go as the following:(1)After the GRAS method is applied,the absolute percentage error(APE)is larger for sectors with smaller to-tal output(or total input),revealing an asymmetry in the update results.(2)By incorporating additional information for sectors with smaller total output(or total input),the TGRAS method produces lower values for the information loss index,information relevance index,standardized total percentage error,and deviation distance index compared to the GRAS method.This effectively mitigates the asymmetry in the GRAS method’s update results,resulting in a significant higher update accuracy for the TGRAS method.(3)Even when the long short-term memory network model is employed to incorporate additional information for sectors with smaller total output(or total input),the TGRAS method shows only a slight increase in the information relevance index,while still main-taining superior update accuracy compared to the GRAS method.
作者
孟彦菊
赵之琛
Meng Yanju;Zhao Zhichen(School of Statistics and Mathematics,Yunnan University of Finance and Economics,Kunming 650221,China)
出处
《统计与决策》
CSSCI
北大核心
2024年第24期48-53,共6页
Statistics & Decision
基金
国家社会科学基金一般项目(20BTJ001)。
