陳思寬臺灣大學：國際企業學研究所石玉禎2007-11-282018-06-292007-11-282018-06-292004http://ntur.lib.ntu.edu.tw//handle/246246/60398自利率自由化以來，我國中央銀行較為重視貨幣總計數的成長，並逐年設定與公布貨幣成長目標區，以便達成對內、對外價格穩定及促進經濟成長。事實上，僅重視貨幣數量的穩定易造成利率的上下波動，而僅重視隔夜拆款利率的穩定易造成貨幣數量的高低起伏，因此，中央銀行在穩定物價的前提下，雖然重視貨幣數量的成長，亦須適時兼顧短期利率的穩定。 在對抗通貨膨脹與景氣衰退時，中央銀行是否有一個簡單的政策法則可以做為貨幣政策的參考與依據，並可同時提供各界解釋與預測短期利率的變動方向？一九九三年由Stanford大學John Taylor教授所提出的「泰勒法則 (Taylor rule) 」，便是一項用以分析和預測美國聯邦資金利率 (Federal funds rate) 走勢的模型，由於該法則之形式簡潔、極具參考價值，因此廣為各界所引用。 泰勒法則認為短期利率之調整，乃是以產出缺口與通貨膨脹率缺口二者為依據，所形成的政策反應函數。當產出與物價的實際值高於目標值時提高利率，反之，當產出與物價的實際值低於目標值時便降低利率；並且，經由短期名目利率的調整，使得穩定實質產出與通貨膨脹兩目標均得以達成。然而在實務上，產出缺口是一個無法實際觀察到的變數，過去文獻大多利用時間序列的方法來做估計，故本文便採用其中的Hodrick-Prescott filter (HP filter) 及Beveridge-Nelson filter (BN filter) 兩種濾波模型對產出缺口加以估計。 本研究嘗試以不同方式對產出缺口進行估計，且將結果帶入泰勒法則以估算出我國隔夜拆款利率，並比較不同濾波模型為基礎的泰勒法則估計結果之差異。最後所得到的結論為：在推估產出缺口方面建議還是採用HP filter法為適。此外，若能將「外生性技術進步趨勢項」因素加以考量並擴充模型，則所得之產出缺口估計值會更為精準，並對於將其代入泰勒法則之結果有所裨益。Since the liberalization of the interest, the central bank of Taiwan puts more emphasis on the growth of monetary quantity. By setting and declaring the target value of it, the central bank of Taiwan can reach the price stabilization and the promotion of economic growth. In fact, only caring for the stabilization of the monetary quantity may lead to the fluctuation of the interest rate, and only minding the stabilization of the interest can result rise and fall in the monetary quantity. Therefore, under the presupposition of price stabilization, the central bank should focus on the importance to not only the growth of monetary quantity but also interest stabilization at the same time. When competing with the inflation and recession, is there a simple policy rule of the central bank to be the reference and basis of monetary policy, and meanwhile to provide the public to explain and predict the variable trend of short-term interest rate? The “Taylor rule” by John Taylor in Stanford University in 1993 is a model to analyze and predict Federal funds rate. Due to the simple and clear form of this rule, as well as its referential value, it leads to a variety of applications in the academic and practical fields. The Taylor rule suggests that the adjustment of short-term interest rate is a function of policy reaction based on the output gap and inflation rate gap. While actual value of output and price is higher than goal value, then interest rate rises, whereas actual value of output and price is lower than goal value, the interest rate lowers. Therefore, through the adjustment of short-term nominal interest rate, two goals of steadying real output and inflation can be reached. However, in practice, output gap is an unobserved variable. Past literature most used time series methods to estimate it. Consequently, in the context I will use two filter models, Hodrick-Prescott filter (HP filter) and Beveridge-Nelson filter (BN filter), to estimate output gap. This research first tries to use different methods to estimate the output gap, and then make use of the results in the Taylor rule to estimate the overnight interest rate in Taiwan, as well as compare with the differences of Taylor rule based on two different filter models. Conclusions as follows: HP filter method is a better way to predict and estimate output gap. In addition, if external technical improvement factor can be taken into consideration and then extend the model, the estimate of output gap derived from the model will be more precise, and will help the results used in the Taylor rule.目 錄 第一章 緒論 …………………………………………………………….1 第一節 研究背景與動機 …………………………….…………………………….1 第二節 研究目的 …………………………………………………………………….3 第三節 研究架構與流程 ……………………………………………….………….4 第二章 理論基礎與文獻回顧 …………………………………………..6 第一節 泰勒法則之簡介 …………………………………….…………………….6 第二節 泰勒法則之文獻回顧 …………………………….………………………9 第三節 產出缺口估計之文獻回顧 ……………………………………………13 第三章 資料分析與變數選取 …………………………………………15 第一節 研究方法 ………………………………………………………………….15 第二節 資料來源與變數選取 ………..……………………………………….23 第四章 實證結果分析 …………………………………………………27 第一節 產出缺口之估計 …………………………………………………………27 第二節 預期通貨膨脹率缺口之估計 …………………………………………38 第三節 泰勒法則實證結果 ……………………………………………………..45 第五章 結論與建議 ……………………………………………………51 參考文獻 表 次 表1 實質GDP之ADF單根檢定 …………………………………………….27 表2 LS法估測dGDP之ARMA模型 ………………………………………..30 表3 Residual tests—Serial Correlation LM Test ………………33 表4 通貨膨脹率之ADF單根檢定 …………………………………………38 表5 LS法估測dInflation之ARMA模型 ..............................40 表6 Residual tests—Serial Correlation LM Test ………………..43 表7 泰勒法則中各變數之ADF單根檢定 ……………………………….45 表8 共整合分析檢定不含時間趨勢項 (共整合含常數項但VAR模型不含常數項) ……………………………………………………………….46 圖 次 圖1 Correlogram of dGDP …………………………………………………29 圖2 Residual tests—Correlogram of Q-statistics ………………31 圖3 Residual tests—Correlogram Squared Residuals ………32 圖4 實質GDP與經BN filter平滑後結果之比較 ………………………34 圖5 實質GDP與經HP filter平滑後結果之比較 ………………………35 圖6 不同平滑法下產出缺口之比較 ……………………………………….37 圖7 Correlogram of dInflation …………………………………………39 圖8 Residual tests—Correlogram of Q-statistics ……………..41圖9 Residual tests—Correlogram Squared Residuals ………42 圖10 預期物價膨脹率缺口 ………………………………………………….44 圖11 我國隔夜拆款利率與以HP filter法 (含外生性技術進步趨勢項) 估計產出缺口之泰勒法則間比較 …………………………..47 圖12 我國隔夜拆款利率與以HP filter法 (不含外生性技術進步趨勢項) 估計產出缺口之泰勒法則間比較 ……………………….48 圖13 我國隔夜拆款利率與以BN filter法估計產出缺口之泰勒法則間比較 …………………………………………………………………….49 圖14 我國隔夜拆款利率與以各種不同方法估計產出缺口之泰勒法則間比較……………………………………………………………………50779315 bytesapplication/pdfen-US泰勒法則BN濾波HP濾波隔夜拆款利率產出缺口估計Taylor ruleoutput gapthe overnight interest rateBeveridge-Nelson filterHodrick-Prescott filter[SDGs]SDG8thesishttp://ntur.lib.ntu.edu.tw/bitstream/246246/60398/1/ntu-93-R91724072-1.pdf