贝叶斯视角下符号约束与时变随机波动SVAR模型的实现与应用

发布时间：2019-02-24 10:51

【摘要】：SVAR模型的传统识别方法更多地涉及等式约束,即通过对模型中的结构参数或脉冲响应函数施加以严格的等式约束,从而实现模型参数的识别与估计,符号约束则是通过将已有经济理论背景转化为先验信息,借助不等式来约束各变量间的关系,能有效地避免模型缺乏理论基础的问题;伴随着经济社会改革的不断深化,模型中变量间系数及扰动项的方差—协方差矩阵均随时间而变,表现出明显的时变特征,固定参数已不能有效地刻画经济运行中的动态变化关系,时变随机波动SVAR模型的优势日益凸显;贝叶斯估计的一个重要特征即是将先验信息与样本信息相结合,应用贝叶斯定理得到参数的后验分布以确定最终估计值,参数估计的精度及可信度大幅度提升。基于此,本文系统地阐述了TVP-SV-SVAR模型的建立及贝叶斯估计过程,将其应用于中国货币政策有效性的量化研究,并就Sign-SVAR、TVP-SVAR及TVP-SV-SVAR三种模型的效果进行了对比分析。实证结果进一步发现,结合贝叶斯估计方法的符号约束与时变随机波动SVAR模型在宏观经济政策问题的考察与分析中更具优势。
[Abstract]:The traditional identification method of SVAR model involves more equality constraints, that is, by imposing strict equality constraints on the structural parameters or impulse response functions in the model, the identification and estimation of the model parameters can be realized. Symbolic constraint can effectively avoid the lack of theoretical foundation by transforming the background of economic theory into prior information and constraining the relations among variables by means of inequality. With the deepening of economic and social reform, the variable-covariance matrix of variables and disturbance terms in the model changes with time, showing obvious time-varying characteristics. Fixed parameters can not effectively describe the dynamic relationship in economic operation, and the advantage of time-varying stochastic fluctuation SVAR model is increasingly prominent. An important feature of Bayesian estimation is the combination of prior information and sample information, and the posterior distribution of parameters is obtained by using Bayesian theorem to determine the final estimation value. The accuracy and reliability of parameter estimation are greatly improved. Based on this, this paper systematically describes the establishment of TVP-SV-SVAR model and Bayesian estimation process, applies it to the quantitative study of the effectiveness of monetary policy in China, and discusses the Sign-SVAR, model. The effects of TVP-SVAR and TVP-SV-SVAR models were compared and analyzed. The empirical results further show that the SVAR model with symbolic constraints and time-varying stochastic fluctuations based on Bayesian estimation method has more advantages in the investigation and analysis of macroeconomic policy problems.
【作者单位】：中央财经大学统计与数学学院;中央财经大学金融学院;中国社会科学院世界经济与政治研究所《世界经济》编辑部;
【基金】：国家社会科学基金重大项目“‘互联网+’推动经济转型机理与对策研究”(15ZDC024) 国家自然科学基金面上项目“货币总量转向信用总量:全球虚拟经济与实体经济背离机理与宏观政策应对”(71473279)资助
【分类号】：O212

【相似文献】