Study On The Trend And Fluctuation Of Rmb Exchange Rate Under The Managed Floating Exchange Rate Regime

July 21, 2005, the People’s Bank of China issued an announcement that China began to implement the reform of the RMB exchange rate. On the principles of initiative, controllability and gradual, China implement managed floating exchange rate regime, based on market supply and demand with reference to a basket of currencies. With exchange rate reform, the RMB began to fluctuate and consistently appreciate. The issue of the RMB exchange rate has become a focus problem concerned by China and the trade partners.The traditional theories about exchange rate are not suitable at present. The managed floating exchange rate system of RMB is different from the floating exchange rate system of developed countries. How to reflect the floating and the management of the exchange rate system, how to effectively grasp the trend and fluctuation of the exchange rate, there need new perspectives and methods to study the exchange rate. In this dissertation, the nonlinear characteristics of the RMB exchange rate series are examined. The nonparametric methods are used to estimate the exchange rate trend. Semi-parametric models reference exchange rate equilibrium theories and intervention theories are constructed to study the change of exchange rate under managed floating exchange rate system. This dissertation mainly includes the following contents.Firstly, I studied the characteristics of the prices series and volatility series of the RMB exchange rate, including non normality, long memory, volatility clustering, asymmetry and so on. The J-B statistic is used to test the non normality of the price series and the volatility series of the exchange rate. The R/S method and the modified R/S method are used to test the long memory of the volatility series of the exchange rate. The GARCH models are used to test the volatility clustering of the exchange rate. The TGARCH model is used to test the asymmetry of the exchange rate. The test results showed that the RMB exchange rate series have the nonlinear characteristics, including “fat tail” distribution, long memory and volatility clustering. This indicated that the traditional financial theories such as efficient market theory and the Brown campaign will no longer be established. The historical informations of the exchange rate can help to predict the trend and the fluctuation of the future exchange rate.Second, nonparametric estimation method was used to study the trend of RMB exchange rate based on the nonlinear characteristics of the series. The nonparametric estimation methods are the GMDH algorithm and the GMDH algorithm based on K nearest neighbor estimation. The GMDH algorithm and the modified GMDH algorithm can minimize the error came from artificial setting. The empirical results show that the nonparametric estimation method can well fit the trend of RMB exchange rate.Third, according to the characteristics of the change of RMB exchange rate, the sample since the reform began to 2013 is divided into three sub samples. GARCH(p, q) models are used to study the fluctuation of RMB exchange rate. The order and the parameters of high order GARCH model are determined by GMDH algorithm. China central bank regulated the range of floating of RMB through foreign exchange intervention. This dissertation studied foreign exchange intervention of RMB under managed floating exchange rate regime. The GARCH model is used to analyze the effectiveness of foreign exchange intervention. The results show that foreign exchange intervention is effective on stabilization fluctuations of the RMB exchange rate.Finally, in order to describe the management factors and the floating of RMB exchange rate, semi-parameter model is constructed to estimate the trend and fluctuation of RMB exchange rate. The economic variables which impact on the exchange rate are chose to constitute the part of parameters according to the equilibrium theories and intervention theories of the exchange rate. The non parametric estimations are used to reflect the rules and characteristics of the exchange rate. The semi-parametric model is LS- kernel estimation. The empirical analysis shows that the predicting effects of LSkernel estimation model are best among these models.