| Summary: | 碩士 === 國立臺北大學 === 統計學系 === 95 === Value at Risk（VaR） is not only a good measurement of market risk for investment portfolio , but also a good way to measure pricing accordingly . In the past , there were many researches and studies on the model of calculating VaR , however , there were less focus on the impact of the regime switching of return series . As a result , the higher estimation of return series variation causes a higher VaR. The more reserve prepared for risk lower the competition of financial institutions.
In this study , the focus is to compare both MAR-GARCH model and Markov switch model modified by Gray（1996）, in order to find out the best way to measure and forecast return series among these two. And further to study the methodology of VaR improvement and application of that specific model identified.
Accordingly to the empirical results, Taiwan 10 year bond interest rate does show the trend of volatility persistence and regime switching. To use those two regimes to describe the return series is proven proper. Although both of MAR-GARCH model and Markov switch model do take into consideration of those two factors, Mar-GARCH is much better in terms of VaR calculation and forecast .
|
|---|
