摘要
中国渐进式的改革实践要求中国宏观时间序列的建模能够允许参数平滑变化,而传统的VAR模型对此无能为力。本文详细阐述了在贝叶斯估计框架下,如何利用MCMC算法,建立时变参数VAR模型的过程,并利用该模型对徐高(2008)的数据重新进行了拟合,发现其文中提出的"斜率之谜"现象不复存在,因此时变参数VAR模型在拟合中国宏观时间序列方面更为精准。
Modeling macro time series in China should allow smooth changes of parameters in the model for Gradual reform practice in China, however, the traditional VAR model cannot afford it. In a Bayesian estimation framework, this paper states a process of building time varying parameters VAR model under MCMC algorithm. Again, we use the model to fit the data in Xugao (2008), and find that the slope puzzle in Xugao no longer exists, so time varying parameters VAR model is more accurate for fitting macro time series in China.
出处
《数量经济技术经济研究》
CSSCI
北大核心
2011年第10期123-133,共11页
Journal of Quantitative & Technological Economics