社会核算矩阵平衡方法研究——最小二乘交叉熵法  

Estimating a Social Accounting Matrix Using LSCE Methods

作　　者：涂涛涛[1] 马强 

机构地区：[1]华中农业大学经济管理学院 [2]湖北农村发展研究中心

出　　处：《数量经济技术经济研究》2012年第7期134-147,共14页Journal of Quantitative & Technological Economics

基　　金：国家社科基金项目"中国特色农业现代化发展之路:农业流通现代化研究"(08AJY020)的资助

摘　　要：针对初始不平衡SAM与真实SAM关系未知的情形,本文提出了最小二乘交叉熵(LSCE)平衡法。基于最小二乘法(LS)、交叉熵法(CE)以及LSCE方法的仿真分析表明,CE与LS的相对稳健性取决于初始不平衡SAM的误差特征:当初始不平衡SAM的交易流量更接近于真实SAM时,LS较优;当初始不平衡SAM的系数矩阵更接近于真实SAM时,CE较优。LSCE方法同时考虑了SAM表流量和系数矩阵信息,故可得到精度介于LS和CE间的平衡SAM表,从而保证了平衡后SAM表的相对精度。In the case of unknown relationship between initial unbalanced SAM and real SAM, this paper proposes a more flexible and robust approach for SAM balancing, namely least squares cross-entropy （LSCE） method. Meanwhile, the relative robustness of the least squares method （LS）, the cross-entropy （CE） and the LSCE method are examined through simulation. The result shows that the rela- tive robustness of CE and LS method depends on the characteristics of error struc- ture in the initial unbalanced SAM. If the flow of the initial unbalanced SAM is closer to the real SAM table, then the LS method is more reliable. On the contrary, if the coefficient matrix of the initial unbalanced SAM is closer to the real SAM table, then the CE method is more reliable. Since the LSCE method takes into account the flow and coefficient matrix information of SAM, the accuracy of the estimated balanced SAM is between LS and CE method, which ensures the robustness of LSCE method.

关 键 词：社会核算矩阵 平衡方法 最小二乘交叉熵法 

分 类 号：F222.1[经济管理—国民经济]