由於使用一些金融之變數（如貨幣供給、貨幣基數、短期利率、借入準備等）代理決策當局的貨幣政策，會產生內生性及不一致性的問題，本文捨棄金融市場中常用的變數，而改採最新流行的敘述性（Narrative Approach）建立貨幣政策指標。接著利用所建立之政策指標估計央行之貨幣政策反應函數。本文將貨幣政策反應函數分為對稱反應函數與非對稱反應函數兩種形式。所謂對稱，即傳統直接型模型，所謂非對稱，指的是物價或產出增加一單位，但央行在可容忍的最高物價上漲率，或預期的最適經濟成長率之上侑之下的反應強度並不相同。 本文之實證結果發現，央行的非對稱行為強列地反應於通貨膨脹、實質產出與景氣虛擬變數，同時央行貨幣政策的取向，明顯受到物價穩定、經濟成長及充分就業三項總體目標變數的影響。因此傳統上將貨幣政策視為外生變數的設定，或直線型設定，並不恰當。

　　　　　Endogenity and inconsistency are two problems that arise when the conventional money market variables, such as money supply, monetary base, and non-borrowed reserve and short interest rates, are used as the proxy for monetary policy. This paper uses the narrative approach to build up a monetary policy index. This index is then employed to examine the central bank reaction function, of which linear and nonlinear forms are considered. Our results show that the monetary policy reacts differently to then inflation rate and real output when economic conditions are different. Strong responses are discovered when the inflation rate and output are above their corresponding tolerance level of the central bank.