摘要
了解经济周期的典型事实是经济周期建模分析的前提,然而周期典型事实对于建模过程中采用的滤波退势方法可能是敏感的。本文以中国主要宏观经济季度数据为研究对象,结合Lomb周期分解方法分析5种常用滤波在提取周期信息方面的特点和经济周期典型事实对滤波退势方法的敏感性。结果显示,变量的交叉相关性对于滤波方法的选择敏感,相对波动性对于滤波方法的选择不敏感,目前数据季节调整配合HP滤波仍是较好的退势方法。
Combined with Lomb periodogram, this paper compares five commonly used filters in extracting business cycles of Chinese seasonal data, and analyzes stylized facts' sensitivity to filtering methods. Results show that relative volatility between variables is not sensitive to filtering method; however, cross-correlation is sensitive to filtering method. Applying HP filter to seasonal adjusted data is still a good choice currently.
出处
《数量经济技术经济研究》
CSSCI
北大核心
2013年第7期131-147,160,共18页
Journal of Quantitative & Technological Economics