基于动态因果结构推断的SVAR模型识别:算法和仿真

发布时间：2018-05-19 21:11

本文选题：SVAR模型 + 模型识别　；参考：《系统工程理论与实践》2016年06期

【摘要】：给定任一满足递归结构假设的SVAR模型,存在一个与其是相同数据生成过程的线性动态因果结构模型,且SVAR模型系数矩阵与动态因果结构之间存在特定的对应关系,故同期变量间因果结构推断可以为SVAR模型提供正确的识别条件.同时还证明同期变量与滞后变量间动态因果结构推断能为同期变量间因果结构的正确推断提供有用信息.据此,在PC算法基础上,构建了动态因果结构推断的具体算法,将基于同期变量因果结构推断的SVAR模型识别拓展到基于动态因果结构推断,从而使SVAR模型得以完全识别的情形得到有效拓展,并给出了该方法下SVAR模型识别的充要条件,这些结论均得到了Monte Carlo仿真结果的有力支持.
[Abstract]:Given any SVAR model that satisfies the hypothesis of recursive structure, there is a linear Dynamic Causality structure model which is the same data generation process, and there is a specific correspondence between the SVAR model coefficient matrix and the Dynamic Causality structure, so the causal structure inference between the same time variables can provide the correct identification conditions for the SVAR model. It also proves that the dynamic causal structure inference between the synchronization variables and the lag variables can provide useful information for the correct inference of the causal structure between the same variables. Based on the PC algorithm, a specific algorithm for dynamic causal structure inference is constructed, and the SVAR model recognition based on the causal structure inference of the same time variable is extended to the dynamic causal structure based on the Dynamic Causality structure. It is inferred that the case of the SVAR model can be fully recognized, and the sufficient and necessary condition for the identification of SVAR model under this method is given. All these conclusions are strongly supported by the results of the Monte Carlo simulation.
【作者单位】：宁波大学商学院;复旦大学管理学院;山东大学经济研究院;
【基金】：国家自然科学基金(71102003,71271058) 国家社科基金(13BJY179) 中国博士后科学基金一等资助项目(2014M560276) 浙江省自然科学基金(LY13G010004)~~
【分类号】：F224

【相似文献】