混合分层结构Gibbs算法与时变因果关系检验及应用

Hybrid Hierarchical Gibbs Algorithm and Time-varying Causality Test and Application

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。

Traditional Granger-causality test method hard to handle the non-stationary data and Timevary causal relationship,so this paper implemented regime Time-varying VAR model with hybrid hierarchical Gibbs algorithm based on infinite state Markov process for Time-vary causality test aimed at the non-stationary data.By empirical analysis of the time-varying characteristics of demand-pull effect in China＇s economic environment,we got the following conclusions：domestic demand is becoming more important to the economic growth;economic growth reduces the dependence on investment and exports;the domestic effect easily be affected by changes of external demand in the transformation process of economic growth,so timely adjustment of monetary policy and driving force of investment are needed to ensure the stability of economic growth,which is good for further upgrading of industrial structure.