次貸風暴前後外匯匯率風險值之比較分析－以美元兌英鎊、歐元、日圓與新台幣為例

A Comparative Analysis of Foreign Exchange Rate on Value at Risk under Sub-Prime-Example of USD against GBP, EUR, JPY and NTD

徐孟資(Meng-Tzu Hsu)

指導教授 ： 許英麟 ； 陳弘吉

朝陽科技大學/管理學院/財務金融系/碩士(2009年)

摘要

外匯市場是世界上最具規模也是流動性最大的金融市場之一，且在資訊爆炸的時代下，各種訊息皆能迅速的被傳遞，因此，外匯市場的匯率變化日益劇烈，投資人不再只是考慮投資報酬率的高低，還有其風險控管的大小，尤其近期金融風暴之影響，使得全球各地金融市場產生巨幅波動，隨著金融市場的劇烈變化，外匯匯率之波動可能更形劇烈，因此，有必要就次貸風暴前後之外匯風險值(Value at Risk, VaR)作一比較，以提供外匯部位之持有者作一個參考。
本研究目的主要探討外匯市場風險值對於次貸風暴前後的問題，運用不同風險值模型(歷史模擬法、蒙地卡羅模擬法、變異數-共變異數模擬法，及ARMA-GARCH模擬法)來計算英鎊、歐元、日元、新台幣與美元在移動窗口30天、60天、125天與250天及99%與95%信賴水準下之風險值比較。
經實證研究發現:
1.各幣別在ARMA-GARCH 95%信賴水準下，穿透率較接近研究所設定的顯著水準，各期間的移動窗口表現較一致。
2.在LRcc檢定與Z檢定的準確檢定中，各幣別以95%信賴水準下所計的變異數-共變異數法與ARMA-GARCH法中的準確性結果較佳。
3.從RMSE效率性檢定來中可以得知，所有模型中，蒙地卡羅法所估計出的RMSE值最小效率性最佳，由99%信賴水準所估計出的RMSE值較小，較具有效率性，在各幣別中則是以台幣所估計出的RMSE結果較佳。
4.由於次貸風暴的發生，金融資產之市場價格波動加劇，各幣別在次貸後所需計提之風險值皆較次貸前為多，其中以英鎊的VaR波動幅度與VaR提列幅度最大。

關鍵字

ARMA-GARCH模擬法 ； 蒙地卡羅模擬法 ； 歷史模擬法 ； 風險值 ； 次貸風暴 ； 變異數-共變異數模擬法

並列摘要

The foreign exchange market is the most mobile and also the largest in the financial markets. At the current information boom, messages can be quickly passed on. Therefore, the foreign exchange rate changes rapidly day by day. The investors not only consider the return level of investments but also its risk management. Especially during the current financial crisis, created massive fluctuations in the global financial market. With the fluctuations, the current exchange rate would have a even more volatile fluctuation. Consequently, it is necessary to analyze the Value at Risk (VaR) of the foreign currency rate before and after the Sub-Prime crises.
This study focuses on the VaR of the foreign exchange market before and after the Sub-Prime crises, using different VaR models (such as historical simulation approach, Monte Carlo simulation approach, variance-covariance simulation approach, ARMA-GARCH simulation approach) to calculate USD against GBP, EUR, JPY and NTD on moving window of 30 days, 60 days, 125 days and 250 days with 99% or 95% confidence level of VaR.
The results are as follows:
1.With the currency set at ARMA-GARCH 95% confidence level, the penetration rate is closer than the set significant level and presented the same performance during each moving window.
2.In the LRcc test and Z test accuracy tests, in which each currency set at a 95% confidence level, the estimated variety value result shows a better outcome in both variance-covariance approach and ARMA-GARCH approach.
3.We can learn from the RMSE test, in all models, the RMSE value is best determined by the Monte Carlo approach, which estimation presents the smallest efficiency, on the other hand, the estimate for RMSE of 99% significant level is of smaller value, and also has more efficiency, amongst different currencies, NTD shows a better result of RMSE estimate.
4.As a result of the Sub-Prime crisis, the fluctuation in the financial market has become more volatile, all currencies have more VaR to consider than before the crisis, amongst them, GBP has the biggest VaR fluctuation rate and range.

並列關鍵字

ARMA-GARCH simulationapproach ； Monte carlo simulation approach ； Historical simulation approach ； Value at risk ； Sub-Prime ； Variance-covariance simulation approach

參考文獻

2.林恩如 (2006)。「風險值衡量方法之匯率實證」。國立成功大學財務金融研究所碩士論文。

1.Alexander, C. O. and Leigh, C. T., (1997), "On the Covariance Matrices Used in Value at Risk Models", The Journal of Derivatives, 4, 50-62.

3.Beder, T. (1995), "VaR： Seductive but Dangerous", Financial Analysts Journal, 51, 12-24.

4.Bollerslev, T., (1986), "Generalized Autoregressive Conditional Heteroscedasticity", Journal of Econometrics, 31, 307-327.

5.Christoffersen, P., (1998), "Evaluating interval forecasts", International Economic Review, 39, 841-862.