结构参数计量经济学模型的识别及其应用  

Identification of Structural Parametric Econometric Models and Its Application

作　　者：白仲林[1] 缪言[1] 韩猛[2] 

机构地区：[1]天津财经大学理学院,天津300222 [2]内蒙古财经大学统计与数学学院,内蒙古呼和浩特010051

出　　处：《数理统计与管理》2016年第3期503-516,共14页Journal of Applied Statistics and Management

基　　金：国家自然科学基金项目(71271142);内蒙古自治区自然基金(2014BS0702)

摘　　要：结构计量模型可识别性决定了结构参数估计的稳定性。为此,本文从扩展结构参数与简化型参数关系体系的视角讨论了结构参数模型的识别问题。首先,证明了一种识别结构参数模型的秩条件。研究发现,对于具有系数线性约束的联立方程组模型,本文的秩条件等价于Koopmans的秩条件;而且,对于SVAR模型,本文的秩条件推广了Hamilton的三角约束条件、Blanchard和Quah的短期识别约束条件、Gali的长期识别约束条件和Rubio-Ramirez等的合并约束条件。另外,应用本文的秩条件研究了一种DSGE模型的可识别性。It is a fact that not all of the structural parameters can be estimated reliably because of identification failure. This paper discusses the sufficient condition of identifiability of the structural parametric model starting from the perspective that extended the correspondent relationships between the structural and reduced-form parameters. At first, a general rank condition of identification is established and proved for a structural parametric model which has certain parametric correspondent relationships. Secondly, the study shows that the rank condition in this paper is equivalent to the Koopmans rank condition for the simultaneous equations model which under the linear coefficient restrictions; the Hamilton triangular restrictions, the short-run and long-run identifying restrictions studied by Blanchard ＆ Quah and Gall and the Rubio-Ramirez co-occurrence restrictions impose on SVAR model which assure the parameters of SVAR model can be identified are particular circumstances of the rank condition theorem in this paper. In addition, the identification methodology is illustrated using a DSGE model.

关 键 词：结构参数模型 结构识别 识别约束 秩条件 

分 类 号：F064.1[经济管理—政治经济学] O212[理学—概率论与数理统计]