自從Clarida, Gali, and Gertler (1998)在理性預期模型的架構下建立對前瞻性泰勒法則的估計方法, 以一般動差估計法(GMM)評估貨幣政策或具前瞻性理性預期模型已成為普遍且熱門的計量方法。本研究的主要目的在檢討以 GMM 估計前瞻性泰勒法則與新凱因斯菲利普曲線的方法, 是否能藉由找出更合適的工具變數而改善其準確度與估計效率。既存實證文獻的工具變數個數相當有限, 這與實務世界中, 央行會同時監測大量經濟與財務數據以形成它對通膨率以及產出預期有所乖離。本研究提出如何根據既有的因子模型來建構更富含預測訊息的因子工具變數, 並評估這種新方法應用於前瞻性理性預期模型估計是否合宜。 ; Since the work by Clarida, Gali, and Gertler (1998) estimating a forward-looking equation in a rational expectation model by the Generalized Method of Moments (GMM) has become a popular approach. Two important issues are discussed. One is estimating a forward-looking Taylor rule; the other one is estimating the New Keynesian Phillips curve. The standard estimation approach dealing both issues is to estimate the model parameters by GMM with a limited number of lagged variables included in the instrument set. However, it is widely agreed that central banks monitor and analyze a wide set of data series which are unlikely to be spanned by just a small number of variables. Therefore, the conventional approach with small instrument set constructed from only a limited number of variables could subject to unappealing small-samples properties of GMM estimators. The key theme of this research is to explore the potential advantages of incorporating richer information set conveyed from a large panel dataset on studying these two issues. We assess whether the proposed factors can better represent the central banker’s (agent’s) expectations on future inflation rates and future output in the Taylor rule (Phillips curve) case and whether the exercise can yield more precise parameter estimates. ; 研究期間 9808 ~ 9907
