基于EEMD-JADE模型的PMI與PPI結(jié)構(gòu)分析及傳導(dǎo)機(jī)制

摘要： 研究目標(biāo):構(gòu)建采購經(jīng)理指數(shù)PMI和生產(chǎn)者價(jià)格指數(shù)PPI結(jié)構(gòu)分量間的傳導(dǎo)機(jī)制。研究方法:使用PMI和PPI同比序列進(jìn)行集合經(jīng)驗(yàn)?zāi)B(tài)分解(EEMD)得到固有模態(tài)序列作為觀察信號(hào),采用特征矩陣聯(lián)合近似對(duì)角化算法(JADE)提取獨(dú)立信號(hào)分量序列并通過游程判定法重構(gòu)出不同頻率的結(jié)構(gòu)分量,最后進(jìn)行Granger因果檢驗(yàn)。研究發(fā)現(xiàn):PMI和PPI重構(gòu)出的高中低頻三個(gè)結(jié)構(gòu)分量分別反映短期波動(dòng)、中期波動(dòng)和長期波動(dòng)。Granger因果檢驗(yàn)表明,高頻分量中PPI和PMI互為因果關(guān)系,傳導(dǎo)時(shí)長為1期;中頻分量中PMI是PPI的因且先行4期;低頻分量PPI是PMI的因且先行10期。研究創(chuàng)新:將時(shí)頻分析方法 EEMD-JADE聯(lián)合算法引入經(jīng)濟(jì)領(lǐng)域。研究價(jià)值:為經(jīng)濟(jì)量化調(diào)控政策提供導(dǎo)向和理論依據(jù)。
Research Objectives:Construct the conduction mechanism between the structure components of PMI and PPI.Research Methods:Decompose PMI and PPI with the ensemble empirical mode decomposition(EEMD)method to obtain fixed base sequences as observed signals,then apply Joint Approximate Diagonalization of Eigen-matrices(JADE)to extract independent signal component sequences,and use run length judgment method to reconstruct structure components of different frequencies.Finally,we take Granger causality test on them.Research Findings:Three structure components of high,medium and low frequency,reconstructed by PMI and PPI,represent the short-term,medium-term and long-term volatility respectively.The Granger causality test shows that,in high frequency components PMIH and PPIH are the interrelationship of cause and effect,and the conduction time is 1month.In medium frequency components only PMIM is the Granger cause of the PPIM with4 phase in advance.And in low frequency components only PPIL is the Granger cause of the PMIL with 10 phase in advance.Research Innovations:Introduce the time frequency analysis method EEMD-JADE into economic field.Research Value:Provide guidance and theoretical basis for the quantitative control policy of economy.