混合分层结构Gibbs算法与时变因果关系检验及应用

发布时间：2019-07-08 15:03

【摘要】：传统的Granger因果检验方法受阻于非平稳数据与时变因果关系的存在,本文以无限状态的Markov过程为基础,设计混合分层结构Gibbs算法,实现区制时变VAR模型,用于非平稳数据的时变因果关系检验。实证分析我国经济环境中,需求拉动效应的时变特征。结果表明:国内需求逐渐成为更加重要的经济增长动力;经济增长降低了对投资和出口的依赖;在增长方式的转变过程中,内需效应易受外需变动的影响而波动,需要适时的货币政策调节和投资效应的推动以保证稳定的经济增长,以利于进一步升级产业结构。
文内图片：

图片说明：图３邋Ｍ２增长率、ＣＰＩ与ＰＰＩ的同比增长率的季度数据逡逑２．４生产成本因素影响下的需求拉动效应逡逑
[Abstract]:The traditional Granger causality test method is hindered by the existence of non-stationary data and time-varying causality. Based on the Markov process of infinite state, the hybrid layered structure Gibbs algorithm is designed, and the time-varying VAR model is realized, and the time-varying causality test for non-stationary data is realized. The time-varying characteristics of demand-pull effect in our country's economic environment are analyzed. The results show that domestic demand is becoming more and more important economic growth power; economic growth has reduced the dependence on investment and export; in the course of the transformation of growth mode, the effect of domestic demand is vulnerable to the change of external demand, It is necessary to push forward the adjustment of the monetary policy and the investment effect in order to guarantee the stable economic growth, so as to facilitate the further upgrading of the industrial structure.
【作者单位】：吉林大学数量经济研究中心;吉林大学商学院;
【基金】：教育部人文社科重点研究基地重大项目(14JJD790043)
【分类号】：F224;F124

【相似文献】