Combination Prediction Method Applied In Forecasting RMB Exchange Rate Against Dollar

Since the disintegrating of Bretton Woods system in 1973, countries in Western implemented floating exchange rate system to replace fixed exchange rate system. The fierce and frequent fluctuation of exchange rate brings huge risk to the governments, financial institutions and international enterprises. Because of the huge risk, expert from China and foreign pay more attention to researching exchange rate risk and how to judge the exchange rate accurately. China improved the RMB exchange rate formation mechanism reform in July, 2005 .RMB exchange rate is no longer pegged solely to the dollar, while the People's Bank of China according to domestic and international economic situation choose several main currencies which would be paid corresponding weight to make a basket of currencies. The People's Bank of China is responsible for domestic and international economic and financial situation, based on market supply and demand. Referring to a basket of currencies to calculate fluctuation of RMB exchange rate. As a result, the risk brought by fluctuation of exchange rate is bigger than before. In this situation, to analyze and forecast the fluctuation of exchange rate are very helpful for the government formulating monetary policy correctly and companies avoiding risk of foreign exchange and the banks controlling risk from foreign exchange. Although RMB exchange rate is no longer pegged to a single dollar, analyze RMB exchange rate against dollar exchange rate will help us to judge RMB exchange rate against other currencies exchange. So we use combination prediction method to analyze the trend of RMB exchange rate against dollar.The theory about forecasting exchange rate before always thought over only one or two factor, so the result is not accurate enough. Referring to the theory of Bates and Granger(1969) ,linear composite model could utilize the advantage of single model effectively and achieve a model better than any single model. In this paper we establish composite model by purchasing power parity model. Minimizing the two model's squared error criterion optimal portfolio weighting coefficients obtained the establishment of forecasting model. The result of composite forecasting model is better than single ones. In the past the array which is forecasted by composite forecasting model is stable, but most exchange rate array is nonstationary. So this paper we use the theory of Wei weixian to establish a composite forecasting model to forecast nonstationary, and empirical analyze the model. The empirical analyze show that the model is better than a single forecasting model.The thesis divides into four parts:The first part is introduction, in this part we introduce the significance, construction and innovation of this thesis, and overview of forecasting thesis of exchange rate.The second part is introductions of the affection factor of exchange rate and exchange rate theory.The third part introduces the thought of linear composite exchange rate forecasting and co integration theory.The fourth part is empirical analysis. In this part we use purchasing power parity model and Monetary model and ARCH model separately to analysis RMB/Dollar exchange rate from Jan, 2001 to Aug, 2006. And establish composite model based on PPP model and ARCH model; Compare the result of three model, analysis the reason of inaccurate.