Research On The RMB Exchange Rate Forecasting Based On Wavelet Analysis

Exchange rate is a convert rate between two countries’currency-Exchange rates play an important role in a country’s economic development. With economic globalization, Exchange rates’influence in economic will grow. After July21,2005the exchange rate reform, the RMB exchange rate gets into a continued volatility stage from a stable stage. Our micro-economies’exchange rate risk awareness is not strong, anti-risk ability is also weak, reserve management and some other issues stand out. So it is necessary to study the RMB exchange rate forecasting method to support the RMB exchange rates’forecasting, and thus provide a reference for solving the problem above.Firstly, this paper tells the research background and theoretical, practical significance and what other people have done in this field with my own evaluating. Exchange rate affected by a variety of different factors and the frequency of exchange rate fluctuations caused by different factors is not all the same (for example, central bank intervention to stabilize the exchange rate, therefore the frequency of exchange rate fluctuations caused by it is small while BOP influenced by external economic conditions when the economic situation is good, the exchange rate fluctuations frequency caused by the influx of foreign currency is greater). This paper analyzes the assumptions and principles of some common model often used to forecast the exchange rate. This paper also analyzes the rationality and function in using those models forecasting exchange rate. Through the analysis we discover wavelet analysis can through using decomposition and single composition turn exchange rate into high and low frequency components and use model to forecast each of them then add up all the forecasting values to get the prediction of the exchange rate. In theoretical if we do like that can improve the prediction accuracy rate of forecasting exchange rate. So we propose the hypothesis that wavelet analysis can improve exchange rates’ forecast accuracy, through comparing the GARCH model and wavelet combined GARCH model method on the RMB/USD exchange rate forecast errors, confirming that wavelet combined GARCH model can indeed improve the prediction accuracy rate of exchange rate. On this basis we use the above modeling process into forecasting RMB/EUR, RMB/JPY for studying if the ability that wavelet combined GARCH model can improve the prediction accuracy rate of exchange rate is inclusive, and the result shows that the method can improve many kinds of exchange rates’ prediction accuracy.In this paper, the empirical analysis generally includes three aspects:1, the original exchange rates’return series through a series of GARCH modeling process to get a group predicted exchange rate;2, using the wavelet analysis to decomposing and single composing the exchange rate and establish GARCH model for each of the component to get their forecasts, and then add up all forecasts to get a group forecast of exchange rate;3,decide the better method by computing and comparing the prediction error.Through the empirical analysis in this paper, we think the reasons why wavelet analysis can improve prediction accuracy of exchange rate as follows:1, Compared to the original exchange rate series, the low frequency from wavelet analysis is approaching the original exchange rate, but it is smoother and it has a high goodness of fitting in building model.2, Wavelet analysis distinguishes the factors that caused exchange rate different fluctuations frequencies, which makes each component more regularity comparing to the original exchange rate series, and improves the predictive ability of GARCH models. This paper also explain the reason why using wavelet into the prediction of RMB/EUR weekly exchange rate improve forecast accuracy only a little.