An Empirical Study Of The Characteristics Of The Rmb Exchange Rate Fluctuations :2005-2009

This paper studied the volatility of the USD/CNY, EUR/CNY, JPY/CNY, HKD/CNY and GBP/CNY exchange rate time series during July 21, 2005, when China exerted the reform of RJVIB exchange rate system, and May 6,2009. With ARCH-LM tests as well as the correlagram of residuals squared test confirmed that the logarithm difference series of the five exchange rate groups do have ARCH effect, I estimated the GARCH and EGARCH models to describe the characteristics of their volatility. I found volatility clustering, persistency, and asymmetric effect in China's foreign exchange market. In view of statistic significance, model constraints and residuals ARCH effect test, I analyzed rationality of the model. GARCH model results show that volatility persistency of the exchange rates is very strong except that of the EUR/CNY series. Since RMB pegged to a basket of currencies including the British pound and US dollar, the GBP/CNY series and the USD/CNY series have the strongest volatility persistency. The HKD/CNY has the weakest volatility persistency. External shocks do not contribute to exchange rate stability, but the external shocks impact the five exchange rate series on different levels. EGARCH model shows that the EUR/CNY and the HKD/CNY do not have significant leverage effects, while the JPY/CNY has the most significant leverage effect.GARCH and EGARCH models for the exchange rate data from July 21, 2005 to April 30, 2009 are unable to completely eliminate ARCH effect of the residuals. However, according to fluctuations in exchange rates, changing the time-series' length and re-estimating GARCH and EGARCH models, we can completely eliminate ARCH effect. This may be due to widely different fluctuations in different time periods, which are caused by the fact that intensity of central bank's intervention on exchange rates varies with time. Therefore, it is impossible to eliminate residuals ARCH effect with the same model for different periods. GARCH model for EUR/CNY can eliminate residuals ARCH effect, but the ARCH coefficient is still insignificant because the Euro's forecast error does not change over time and depend on the previous value as much as the US dollar and other foreign currency do. This is because the EUR/CNY exchange rate market is not easily influenced by rumors, political changes and so on. The JPY/CNY model still has residuals ARCH effect. This may be due to its forecast error's variance over time and dependence on previous values are very strong, so even if we narrow sample size, we still can not find a fit model to eliminate the ARCH effect of residuals.Theil inequality coefficient (U) showed unsatisfactory results. This stems mainly from the variance of the forecasted values rather than from the forecast error of the system. Therefore, the reason why ARCH models fit the international foreign exchange market well is that the international foreign exchange market is too huge f to manipulate. However, the RMB exchange rate is under the country's strict controls, so that its volatility is difficult to be accurately portrayed. This requires the implementation of the reform of RMB exchange rate regime. We should gradually make the RMB exchange rate to float freely so as to better avoid exchange rate risks.