GRAS方法的改进及对比研究——基于社会核算矩阵调平和投入产出表更新  被引量:10

A Comparative Study on the Improvement of the GRAS Method for Updating the Social Accounting Matrix and Input-Output Table

在线阅读下载全文

作  者:何志强[1] 刘兰娟[1] He Zhiqiang;Liu Lanjuan(School of Information Management&Engineering, Shanghai University of Finance and Economics)

机构地区:[1]上海财经大学信息管理与工程学院

出  处:《数量经济技术经济研究》2018年第11期142-161,共20页Journal of Quantitative & Technological Economics

基  金:教育部人文社会科学研究项目"提高劳动报酬占比的理论;实证与政策:基于CGE模型的研究"(12YJA790086)的资助

摘  要:研究目标:改进GRAS方法,减少社会核算矩阵(SAM表)调平和投入产出表(IO表)更新的先验信息丢失。研究方法:对比GRAS及其改进的AGRAS、UGRAS、SGRAS、IGRAS等方法,针对保零、保凸、保号、无偏等特点,构建EGRAS方法,并进行了数值模拟。研究发现:综合AIL、Theil’s U、GDM等误差指数,EGRAS方法具有保号、保凸、保零、无偏、避免正负项抵消等优点,并能有效保留先验信息,有推广价值。研究创新:构建保号、保凸、保零、无偏、避免正负项抵消的EGRAS方法进行SAM表调平和IO表更新,并无须定义0/0和0ln0值,而且可求全局最优解。研究价值:在SAM表调平和IO表更新时,减少先验信息丢失,提高调平的经济学含义。Research Objectives:This paper effectively develops the methods of updating social accounting matrix and input-output table and minimizes information loss.Research Methods:The EGRAS method is constructed to compare with the GRAS method including the AGRAS、UGRAS、SGRAS and IGRAS.Then the method is applied to numerical simulation.Research Findings:Either calculation errors are made or different information gain measures are compared.The numerical results show that EGRAS which is unbiased and convex consistently outperforms both sign-preserving and strictly zero-preserving when increasing and decreasing cell values occur together.Moreover,EGRAS can overweigh large errors in small coefficients.Research Innovations:The research improves GRAS and forms the optimal model-EGRAS which unnecessarily defines 0/0 and 0ln0.Moreover,the global optimal solution can be found.Research Value:It s a core objective to decrease information loss and increase economic importance through EGRAS when updating the Social Accounting Matrix and Input-Output Table.

关 键 词:GRAS 超越方程 保号 保零 社会核算矩阵 投入产出表 

分 类 号:F222.1[经济管理—国民经济]

 

参考文献:

正在载入数据...

 

二级参考文献:

正在载入数据...

 

耦合文献:

正在载入数据...

 

引证文献:

正在载入数据...

 

二级引证文献:

正在载入数据...

 

同被引文献:

正在载入数据...

 

相关期刊文献:

正在载入数据...

相关的主题
相关的作者对象
相关的机构对象