Study On Exchange Market's Volatility And Calculation Of VaR Based On Fast Diffusion Process

Study on foreign exchange's volatility and measurement of risk are the key point of exchange risk management. Nowadays our country is carring out the rate-forming mechanism of RMB reform, so it is very signifcant for effective prevention of foreign exchange risk and even success of the foreign exchange mechanism reform to research the foreign exchange market's volatility.This paper introduce a Continuous Time Stochastic Model(CTSM) with a statistic distribution term. This model is characterized by adding a feedback factor determined by fast diffusion process to the volatility term of geometry Brown motion model. Compared with discrete time model, like GARCH, SV model, the model deals with foreign exchange market's volatility on the market dynamic method. Compared with usual geometry Brown motion model, this model introduce volatility feedback to modify the assumption that volatility rate is a constant of geometry Brown motion model, showing financial time series'high-peak, fat-tail and long term memory character. On this basis, estimating method of model's key parameter q is improved and individual models in different time frequency of Euro/USD exhange rate and other sample exchange rates are established, using high frequency datas. For AUD/JPY exchange rate which has no ARCH effect, q parameter in CTSM is conspicuous.This paper uses Monte Carlo simulation to calculate daily VaR to measure foreign exchange risk. It is shown by VaR accuracy test that the high the time frequency is , the better calculated VaR performs in covering actual lost. All VaRs in 5 minutes time frequency models get through the accuracy test. It is also found by comparing that CTSM based on fast diffusion process is more suitable to describe high frequency foreign exchange datas than CARCH-VaR model.