论文部分内容阅读
研究目标:构建采购经理指数PMI和生产者价格指数PPI结构分量间的传导机制。研究方法:使用PMI和PPI同比序列进行集合经验模态分解(EEMD)得到固有模态序列作为观察信号,采用特征矩阵联合近似对角化算法(JADE)提取独立信号分量序列并通过游程判定法重构出不同频率的结构分量,最后进行Granger因果检验。研究发现:PMI和PPI重构出的高中低频三个结构分量分别反映短期波动、中期波动和长期波动。Granger因果检验表明,高频分量中PPI和PMI互为因果关系,传导时长为1期;中频分量中PMI是PPI的因且先行4期;低频分量PPI是PMI的因且先行10期。研究创新:将时频分析方法 EEMD-JADE联合算法引入经济领域。研究价值:为经济量化调控政策提供导向和理论依据。
Research Objectives: To construct the transmission mechanism between PMI and PPI structural components of PMI. Research Methods: Using the empirical mode decomposition (EEMD) of PMI and PPI year-by-year sequences to obtain the intrinsic mode sequence as the observation signal, the independent signal component sequence is extracted by using the feature matrix joint approximation diagonalization algorithm (JADE) Structure components of different frequencies, and finally Granger causality test. It is found that the three components of high, medium and low frequency reconstructed by PMI and PPI reflect short-term, medium-term and long-term fluctuations, respectively. Granger causality test showed that the PPI and PMI in high-frequency components are causally related to each other, and the duration of conduction is one. The middle-frequency component is PPI due to the first four periods and the low-frequency component PPI is PMI due to the first ten periods. Research innovation: The time-frequency analysis method EEMD-JADE joint algorithm into the economic field. Research value: provide guidance and theoretical basis for economic quantitative regulation and control policy.