| Summary: | 碩士 === 國立臺灣大學 === 財務金融學研究所 === 88 === Abstract:
In U.S.A, Mortgage-Backed Securities have developed for more than two decades and play an influential role in the capital market. The development of MBS market not only helps to improve the liquidity risk of financial institutions but also provides another investitive tool for institutional investors. Therefore, the real estate investment trust is put into action now in the domestic market.
This research attempts to apply Value-at-Risk model to measure market risk of MBS, which furthers investors'' understanding of securitization and promotes development of MBS market. On the basis of RiskMetrics model and Monte Carlo model, this research uses equal weighted moving average method and exponential weighted moving average method to assess the first-order and second-order VaR of several MBS products in the U.S. capital market, such as Mortgage Passthrough securities、PO、IO、Planned Amortization Class bond and Companion class which are compared with Amortizing level payment bond. The conclusion is as follows:
1. On the assumption of normal distribution, using exponential weighted moving average method to estimate standard deviation is able to reflect the trend and volatility of yield curve better. The method is worthy to be proposed to investors.
2. In this research, the contribution to VaR from the second approximation is extremely small, which implies that the first approximation method could catch most of market risk in the U.S. capital market.
3. VaR can correctly express the character of cashflow payment and risk exposure of MBS. For the MBS products in this research, Companion class exposes the most risk and IO exposes the least risk. Besides, the VaR estimates of all above MBS products are lower than that of Amortizing level payment bond.
|
|---|
