指導教授：蘇永成臺灣大學：財務金融學研究所鍾秉諺Chung, Ping-YenPing-YenChung2014-11-272018-07-092014-11-272018-07-092013http://ntur.lib.ntu.edu.tw//handle/246246/262498本研究之最終目標為找出能最適配置美國量化寬鬆期間美元兌新台幣連續報酬率之模型。本研究使用對稱之GARCHM與兩種不對稱之GARCHM模型:一種為GJR-GARCHM(旋轉效果)，另一種為NA-GARCHM(平移效果)。本研究亦將AR(1), MA(1),或ARMA(1,1) 放入條件平均數式中。本研究得到在99%和95%信心水準下之VaR預測，並進行穿透分析，以檢驗各GARCHM模型於市場風險管理之能力與資本準備之效率。主要之發現如下： (1) 若由99%信心水準下之穿透次數來判斷，對稱之GARCHM模型最少，但GJR-GARCHM模型幾乎和對稱之GARCHM模型一樣好，而 NA-GARCHM模型最差。 (2) 考慮絕對平均VaR時，對稱之GARCHM模型的值最大，GJR-GARCHM模型居中，而 NA-GARCHM模型的值最小。較大之絕對平均VaR意指較少之穿透。 較少之穿透較安全，但其代價為較嚴格之VaR與較多之資本準備。與對稱之GARCHM模型相比，GJR-GARCHM模型有多一些之穿透但少一些之絕對平均VaR，即其要求較少之資本準備，但其穿透數僅增加些許，且仍在Basel之允許範圍內。故GJR-GARCHM模型能最適配置美國量化寬鬆期間美元兌新台幣連續報酬率且最有效率。 (3) 在99%信心水準下，MA(1)-GJR-GARCHM(1,1)為最適模型，因其有最小之絕對總合穿透與絕對平均穿透。在95%信心水準下，GJR-GARCHM(1,1)為最適模型。然而，在99%信心水準下，最原始之GJR-GARCHM(1,1)模型仍有用且對稱之GARCHM模型仍具有參考價值。 (4) 在本研究中有兩個例外：AR(1)-NA-GARCHM(1,1)和ARMA(1,1)-NA-GARCHM(1,1)。其結果常異於其他模型。在本研究中NA-GARCHM最差。 (5) 在本研究中，於條件平均數式中加入AR(1)和ARMA(1,1)，會造成穿透次數、絕對總合、絕對平均與絕對最大穿透數增加且造成絕對平均VaR減少。可能之原因為臺灣央行會某種程度上控制NTD/USD，且臺灣央行可能故意干預使NTD/USD市場失去部分可預測性。臺灣央行之干預可能為AR(1)無法完全有效衡量兩期間NTD/USD報酬率之序列相關的原因。The ultimate goal of this research is to find out which model fits the NTD/USD continuous rate of return in the U.S. QE period data best. Symmetric GARCHM and two asymmetric GARCHM models are utilized: one is GJR-GARCHM (for rotation) and the other is NA-GARCHM (for shift). AR(1), MA(1), or ARMA(1,1) is taken into the conditional mean equation. VaR forecasts under 99% and 95% confidence level are obtained and the forward test with violation analysis is done to examine the capability in market risk management and efficiency in capital reserves of each GARCHM model. The main discoveries are as follows: (i) If judged from the violation numbers under 99% confidence level, symmetric GARCHM models have the fewest, but GJR-GARCHM models are almost as good as symmetric GARCHM, and NA-GARCHM models are the worst. (ii) Considering the absolute mean VaR, symmetric GARCHM models have the largest value, GJR-GARCHM models are in the middle, and NA-GARCHM models have the smallest. Larger absolute mean VaR value means fewer violations. Fewer violations are safer, but the price is the stricter VaR and more capital reserves. Compared with symmetric GARCHM models, GJR-GARCHM models have somewhat more violations but somewhat less absolute mean VaR, which means fewer capital reserves are required in GJR-GARCHM models but their violations just increase a little bit and still in the allowance range of Basel Accord. Therefore, GJR-GARCHM models fit the NTD/USD continuous rate of return data in the U.S. QE period best and most efficiently. (iii) Because of the smallest absolute aggregate violation and absolute mean violation, MA(1)-GJR-GARCHM(1,1) is the best under 99% confidence level. GJR-GARCHM(1,1) is the best under 95% confidence level. However, the original GJR-GARCHM(1,1) model is still useful under 99% confidence level and symmetric GARCHM models are still full of reference values. (iv) In this research, there are two exceptions: AR(1)-NA-GARCHM(1,1) and ARMA(1,1)-NA-GARCHM(1,1). Their results are often different from others. NA-GARCHM models are the worst for the data in this research. (v) In this research, the involvement of AR(1) and ARMA(1,1) in the conditional mean equation causes the violation number, absolute aggregate, absolute mean, and absolute maximum violation to increase and absolute mean VaR to decrease. The possible reasons may be that NTD/USD is controlled in some degree by the Central Bank of Taiwan, and the Central Bank of Taiwan may intentionally does some interventions to make the NTD/USD market somewhat out of order and out of forecast ability. The interventions by the Central Bank of Taiwan may be the reason that the AR(1) is not capable of being fully effective in measuring the two-period serial correlation of Rt of the NTD/USD.Section I Introduction…9 1.1 An Introduction to Quantitative Easing (QE)…9 1.2 An Introduction to Value at Risk (VaR)…12 1.3 Motivation and Purposes…14 1.4 Framework…17 Section II The Basel Accord…18 2.1 The Basel Committee…18 2.2 The Basel I Accord (1988)…18 2.3 The Amendment (1996)…19 2.4 The Basel II Accord (2004)…21 2.5 The Basel III Accord…22 Section III Literature Review…25 3.1 Linsmeier and Pearson (2000)…25 3.2 Hull and White (1998)…26 3.3 Berkowitz and O’Brien (2002)…27 3.4 Hsieh (1988)…29 3.5 Bollerslev, Chou, and Kroner (1992)…30 Section IV Data…32 4.1 Source and Processing…32 4.2 An Overview of NTD/USD…33 4.3 An Overview of Continuous Rate of Return of NTD/USD in Percentage…34 Section V Methodology…37 5.1 The Procedures…37 5.2 GARCHM(1,1)…38 5.3 GJR-GARCHM(1,1)…40 5.4 NA-GARCHM(1,1)…41 Section VI Empirical Results…43 6.1 Model Robustness…43 6.2 Parameter Estimates…43 6.3 Explanation to VaR Figures…48 6.4 Out-of-Sample Forward Test with Violation Analysis…49 6.5 Discussion…55 Section VII Conclusion…61 References…65 Figure 1: NTD/USD (2008/11/25~2012/12/28)…69 Figure 2: The Distribution of NTD/USD (2008/11/25~2012/12/28)…69 Figure 3: Continuous Rate of Return of NTD/USD in percentage (2008/11/26~2012/12/28)…70 Figure 4: The Distribution of Continuous Rate of Return of NTD/USD in Percentage (2008/11/26~2012/12/28)…70 Figure 5: Return and VaR in GARCHM(1,1)…71 Figure 6: Return and VaR in AR(1)-GARCHM(1,1)…71 Figure 7: Return and VaR in MA(1)-GARCHM(1,1)…72 Figure 8: Return and VaR in ARMA(1,1)-GARCHM(1,1)…72 Figure 9: Return and VaR in GJR-GARCHM(1,1)…73 Figure 10: Return and VaR in AR(1)-GJR-GARCHM(1,1)…73 Figure 11: Return and VaR in MA(1)-GJR-GARCHM(1,1)…74 Figure 12: Return and VaR in ARMA(1,1)-GJR-GARCHM(1,1)…74 Figure 13: Return and VaR in NA-GARCHM(1,1)…75 Figure 14: Return and VaR in AR(1)-NA-GARCHM(1,1)…75 Figure 15: Return and VaR in MA(1)-NA-GARCHM(1,1)…76 Figure 16: Return and VaR in ARMA(1,1)-NA-GARCHM(1,1)…76 Figure 17: Comparison of All Models Under 99% Confidence Level…77 Figure 18: Comparison of All Models Under 95% Confidence Level…78 Table 1: Summary Statistics of NTD/USD (2008/11/25~2012/12/28)…79 Table 2: Summary Statistics of Continuous Rate of Return of NTD/USD in Percentage (2008/11/26~2012/12/28)…79 Table 3: Likelihood Ratio Test…80 Table 4: Parameters Estimated in GARCHM(1,1)…83 Table 5: Parameters Estimated in AR(1)-GARCHM(1,1)…83 Table 6: Parameters Estimated in MA(1)-GARCHM(1,1)…84 Table 7: Parameters Estimated in ARMA(1,1)-GARCHM(1,1)…84 Table 8: Parameters Estimated in GJR-GARCHM(1,1)…85 Table 9: Parameters Estimated in AR(1)-GJR-GARCHM(1,1)…85 Table 10: Parameters Estimated in MA(1)-GJR-GARCHM(1,1)…86 Table 11: Parameters Estimated in ARMA(1,1)-GJR-GARCHM(1,1)…86 Table 12: Parameters Estimated in NA-GARCHM(1,1)…87 Table 13: Parameters Estimated in AR(1)-NA-GARCHM(1,1)…87 Table 14: Parameters Estimated in MA(1)-NA-GARCHM(1,1)…88 Table 15: Parameters Estimated in ARMA(1,1)-NA-GARCHM(1,1)…88 Table 16: Violation Number Allowed in Basel Accord…89 Table 17: Violation Number and Mean VaR in GARCH Models Under 99% Confidence Level…89 Table 18: Violation Number and Mean VaR in GARCH Models Under 95% Confidence Level…90 Table 19: Model Comparison Under 99% Confidence Level…91 Table 20: Model Comparison Under 95% Confidence Level…924824382 bytesapplication/pdf論文公開時間：2019/07/15論文使用權限：同意有償授權(權利金給回饋學校)量化寬鬆風險值一般化自我回歸條件變異數模型巴塞爾協定自我回歸移動平均穿透thesishttp://ntur.lib.ntu.edu.tw/bitstream/246246/262498/1/ntu-102-R00723004-1.pdf