Based On The Rmb Exchange Rate Density Forecast Combination Forecast

Widely carried out in current economic activities, study on foreignexchange risk attracts more and more attention. How to accurately predictthe change trend and gradient of the exchange rate becomes the basis offoreign exchange risk management. Therefore, it is quite meaningful forthe macroeconomic study of a country and foreign exchange riskmanagement to learn and predict the dynamic behavior of the exchangerate.The real exchange rate of RMB is chosen as the objective of thisstudy. Firstly, the history of RMB development is reviewed and the recenthot issue that the international society requires RMB appreciated is alsodiscussed. The properties of RMB's real exchange rate are analyzed.Because the density forecasting provides a full description of theuncertainty of a random variable at some future time and becomes auseful method of analyzing the uncertainty of forecasting results infinancial institutions of many countries, this paper systemicallyintroduces this kind of forecasting method. According to analyzing theproperties of behavior of RMB's real exchange rate, the nonlinear regimeswitch model——smooth transition autoregressive model and thegeneralized autoregressive conditional heteroscedasticity model areselected as two single-method forecasts to describe the RMB's realexchange rate behavior. These estimated models are used to forecastRMB's real exchange rate out of.sample. Then, the nonlinear combinationforecasting model relying on artificial neural network methodology isproposed, applied to density forecasting of real exchange rate finally.All outcomes imply that the asymmetric and nonlinear dynamicbehavior of RMB's real exchange rate is exhibited, and the combinationforecasting model produces more satisfactory density forecasts ofexchange rates than individual ones.