Threshold Autoregressive (TAR) Model And Study It In The Exchange Rate Fluctuations

As the economic development, social progress, gradually raise the living standards of people, and globe economic pattern changes, makes the exchange rate constantly changing. Due to the series of sudden change factors of Asian financial crisis in1997, China’s entry to WTO in2001, reforms of RMB exchange rate in2005, and globe economic crisis in2008, make the RMB exchange rate against the U.S. dollar presents sudden change situation. Because the changes of exchange rate data are mostly sub-linear, so the traditional linear time series model has its own shortcomings, which is could not effectively reflect the nonlinear nature of economic variables. This paper uses linear time series model and nonlinear time series with threshold auto-regressive model on RMB/dollar exchange rate respectively for fitting, comparing and analyzing the error. The result shows that threshold auto-regressive is more ideal than linear time series model in predicting economic variables.This paper consists of three parts:the first part is mainly introduced the development process and the influence of RMB, and analyzed the main factors of changes in the exchange rate; the second part is introduced the development and application scope of linear time series model and non-linear threshold auto-regressive TAR model; the third part is made use of the method which is discussed previously on changes of RMB/dollar exchange rate for model fitting, forecasting analysis, and comparing of all sorts of model fitting and predict error. The results show that threshold auto-regressive model in explaining economic variables is a more ideal model.