| How to properly analyze and predict the future exchange rate movements has become a focus of research economists, the successful prediction exchange rate movements on the Government to formulate appropriate monetary policy, to avoid foreign exchange risk plays a crucial role. Therefore, this article is a study on exchange rate forecasts, to find a suitable model to predict the trend of the RMB exchange rate. Since July 21,2005, China began to reform the exchange rate, to give up the fixed exchange rate regime pegged to the dollar, the implementation of the basis of market supply and demand with reference to a basket of currencies, a managed floating exchange rate system. We can change after the Chinese exchange rate movements, is divided into three stages. The first stage is from the reform beginning to the financial crisis. The second stage is from the financial crisis to June 19,2010. As the financial crisis, China began to re-imposed exchange controls, long-term RMB exchange rate stable at a relatively constant level. Such a policy is due to unstable economic conditions abroad, in order to prevent the transmission of foreign shocks to China, but also to give domestic firms a stable business environment. Therefore, the exchange rate temporarily slowed down the reform process, implementation of controls on the exchange rate. The third stage, June 20,2010 until now. Little stability in the financial crisis, the Chinese measure of the economic situation at home and abroad, in order to solve the economic problem of internal and external imbalances, restart the RMB exchange rate reform process, the increasing appreciation of the RMB. As time progresses, an increasing rate of RMB exchange rate, volatility is also growing. China's economy and enterprises are facing increasing pressure. Due to the time series model requires data consistency, that is, its internal generation mechanism is the same, so from 2005 to 2010, the RMB exchange rate data can not be fully used. In particular, among which the second phase of the RMB exchange rate under control, did not change basically. The data at this stage is no research value. Therefore, this study of selected recently from now the third stage, the interval is 20 June 2010 to December 31,2010 the daily data, a total of 195 sample points.The exchange rate is an important area of international finance policy tools, both inside and outside the national economy play a regulating role of leverage and also maintain the currency exchange with other countries and bridge ties. Therefore, the RMB exchange rate has been the focus of attention of scholars home and abroad issues. With the birth of a variety of calculation methods various software development, the scholars on the exchange rate to predict the outcome widely predicted to continuously improve the accuracy. Research methods on the prediction generally in three ways:first is the basic factoring in forecasting method, and second, time series analysis and the third is data mining. This paper studies the time series model with a variety of short-term exchange rate expectations, to evaluate the effect of forecast model to evaluate the advantages and disadvantages of each model place. As the RMB exchange rate series is non stationary series, need to enter after the first difference time series model (the symbol is too professional, change the text description). Therefore, we first select the ARIMA model was to model predictions. High-frequency sequences, taking into account the financial heteroscedasticity that often appear in the situation, in ARIMA (1,1,1) on the basis, we carried ARIMA-GARCH (1,1) modeling to short-term RMB exchange rate forecast found that ARIMA-GARCH model relative to the ARTMA model to predict the average absolute deviation (this concept was too vague, change the average absolute deviation) increased by 14%, forecast standard deviation (standard deviation of change predicted) decreased (reduced or reduced) by 12%. On this basis, I take into account the RMB exchange rate reform, despite the increased volatility of exchange rates, but basically is in the process of a revaluation, so there are definitely asymmetric, the RMB exchange rate of the ups and downs of their sequence is inconsistent. Therefore, we establish a threshold autoregressive model (TAR) to capture the exchange rate time series asymmetric, in order to improve forecast accuracy. The results show that the threshold regression model can greatly improve forecast accuracy and stability. In the ARIMA-GARCH model, based on the prediction accuracy increased 41%,32% improved stability. Therefore, among the three models, TAR model predictive ability of the strongest, ARIMA-GARCH model is followed, ARIMA model predictive ability of the weakest. I have been thinking, combined with previous use of EGARCH model to predict the effect has also been a better experience, the reason that the TAR model forecasts better, in its full account of the appreciation of the RMB exchange rate during the non-symmetry.The innovation of this paper is the use of multiple models to predict exchange rate, and compare the various models in predicting the merits of the RMB exchange rate. Estimates of all models are realized in the SAS system, all the programming codes are given in the appendix. At the same time, this fully into account the rise in the RMB exchange rate basically is in the process, with non-symmetry, using the threshold autoregressive model (TAR), greatly improve the prediction accuracy and stability. |
PDF Full Text Request