- 學位論文
貨幣政 策與痛苦指數雙變數的長短期因果非線性關係探討
The Study on the Nonlinear Dynamic Causal Relationships between Monetary Policy and Two Variables of Misery Index
摘要
本研究運用Enders and Granger (1998)和Enders and Siklos (2001)所發展出的動能門檻自我迴歸模型(Momentum-Threshold Autoregressive Model, M-TAR)及動能門檻誤差修正模型(Threshold Error-Correction Model, TECM)，探討金融海嘯發生前後，國家施政的「貨幣政策」與總體經濟雙惡魔變數的「通貨膨脹」與「失業率」之長短期非線性因果關係，並以美國聯邦準備會理事長柏南克(Bernanke)意識到金融海嘯發生之嚴重性而啟動第一次升息的2007年9月18日的月資料時點(9月)做為分界點，進行金融海嘯發生前後變數間之因果關係是否兩期間產生變化之比較分析。 於門檻共整合檢定中發現，貨幣政策與通膨率間互動於金融海嘯前及金融海嘯後及貨幣政策與失業率間於金融海嘯之前及之後，四者之最適模型設定分別為M-TAR、TART、M-TAR及M-TART模型，而此四個門檻模型設定之門檻值 分別為-0.15872、-0.28000、-0.11493及-0.14773。並發現貨幣政策無論是與物價通膨之間，或是與失業率之間，也無論期間是在金融海嘯前或金融海嘯後，皆存在有長期均衡趨勢的非對稱共整合關係。 最後，綜合本研究門檻誤差修正模型因果關係檢定之結果，可得知貨幣政策與通膨率間及貨幣政策與失業率間之短期因果，無論於金融海嘯前或金融海嘯後，其間「領先-落後」之短期因果皆不存在；然而，其間「領先-落後」之長期因果卻因金融海嘯之發生而有所變化。
並列摘要
The aims of this research is to employ nonlinear methods of momentum-threshold autoregressive model (M-TAR) and threshold error-correction model (TECM) elaborated by Enders and Granger (1998) and Enders and Siklos (2001) to investigate the nonlinear dynamic causal relationships between monetary policy and each of the two variables of misery index, inflation rate and unemployment rate during the subprime crisis period. For comparison, we divide the subprime crisis into two sub-period, pre- and under-, using the date of September 18, 2007, the first cut of the Federal fund rate from 5.25% to 4.75% by the chair of the FED (Bernanke), as the divider. The first findings from the threshold cointegration is that no matter it is in the pre-crisis or the under-crisis period, they all show the existence of the asymmetric cointegration relationships between monetary policy and each of the two variables of misery index. For our model specification, with the exception that the relationships between monetary policy and the inflation rate during the under-crisis period is shown as TAR model, all others are shown as M-TAR models. Nonetheless, the threshold values are found to be -0.15872, -0.28000, -0.11493 and -0.14773 for the four models specified. Finally, the overall findings from our TECM examination illustrate that monetary policy and each of the two variables of misery index do not show any short-run causal relation. However, for the long-run causal relation, the lead-lag relationships between monetary policy and inflation rate and between monetary policy and unemployment rate both changed from pre-crisis period to under-crisis period.