The Prediction Of Volatility Of Exchange Rate Based On Median GARCH Model

The CNY exchange rate has encountered distinct changes since Dec. 21.2005, when the reform of CNY exchange brought out. Nowadays, the volatility has more and more frequently and more violent. So the accurate prediction of the exchange rate volatility is necessary.The main contents of this dissertation are:1. Exploration of the feature of exchange rate as a time series data, such as normality, nearly-zero-average, distribution with biased, etc. Four groups of data, CNY/USD and CNY/EUR and CNY/JPY and CNY/HKD, were studied in this paper. Also the change of the four groups of data before and after the exchange rate reform is investigated. 2. Introduction of the ARCH model, GARCH model, quantile regression model, AIC criterion, Mc Leod. Li test method. 3. The study of the four groups of data upon three window widths as 1000, 500 and 200 respectively, the difference between different window widths are illustrated. In each case, the auto-correlation, partial auto-correlation and Mc Leod. Li-testing are exploited. 4. A new model, median GARCH model, based on GARCH and quantile regression, is proposed and the performance is illustrated by using the four groups of data. From the prediction error we can see that the median GARCH model can reduce the prediction error distinguish.