永利集团3044官网欢迎您
WISE
Chow Institute
User Center
Old Version
Chinese
CHINESE
EN
About
About Us
Contact US
People
Faculty Directory
Staff Dirctiory
Program
The Undergraduate Program
The Graduate Program
International Graduate Program
The Part-time Graduate Course Program
Study Aboad Program
Course Catalogue
Research
Publications
Working Papers
News & Events
News
Announcements
Academic Calendar
Seminars
Workshop & Conference
EN
CHINESE
EN
WISE
Chow Institute
User Center
About
About Us
Contact US
People
Faculty Directory
Staff Dirctiory
Program
The Undergraduate Program
The Graduate Program
International Graduate Program
The Part-time Graduate Course Program
Study Aboad Program
Course Catalogue
Research
Publications
Working Papers
News & Events
News
Announcements
Academic Calendar
Seminars
Workshop & Conference
Research
Research
Publications
Working Papers
Research
Publications
Working Papers
Working Papers
Location:
Home
->
Research
->
Working Papers
-> Content
基于贝叶斯模型平均 (BMA) 方法的中国通货膨胀的建模及预测
id: 2207
Date: 20131205
status: published
Times:
Author
CHEN Wei, NIU Linlin
Content
模型和参数的不确定性、以及信息的综合有效利用是影响宏观变量预测精度的主要因素。本文运用贝叶斯模型平均(BMA)方法建模并对样本外通胀进行预测,综合备选模型及变量的信息,以控制模型不确定性,并有效利用丰富的宏观数据信息。本文选取28个解释变量构建了含有2^28个单一线性模型的集合,实证上采用了马尔科夫链蒙特卡洛模型综合算法(MC^3)对备选模型进行抽签,抽签次数为1000万次。采用中国宏观数据的实证结果表明,通胀一阶滞后项与工业企业增加值增速作为预测因子几乎被选择在所有预测模型中;对于通胀的样本内拟合,贝叶斯模型平均(BMA)方法优于单一模型;对于样本外预测,在 标准下,贝叶斯模型平均方法的预测能力优于较为流行的 模型、主成分分析模型、菲利普斯曲线模型、利率期限结构模型、单一最优模型和五变量模型。
JEL-Codes
E31;E47;C11
Keywords
贝叶斯模型平均;通货膨胀;蒙特卡洛模拟;MC^3
TOP