Nonlinear Analysis Of Chinese And Malaysian Exchange Rates Predictability With Monetary Fundamentals

The Chinese Renminbi (RMB) and Malaysian Ringgit (MYR) are pegged to US Dollar during the 1997-98 Asian financial crisis and continued up until the China and Malaysia de-pegged their currencies and announced a new exchange rate regime at the same day on 21st July 2005. By focusing on the post-July 2005 period (August 2005 to July 2010), this paper study the predictability of monthly RMB/USD and MYR/USD exchange rates in different forecast horizons using the generalized regression neural network (GRNN) with discrete and relative monetary fundamentals. Based on a random validation set, the optimal smoothing parameter in the GRNN is attained and subsequently utilized in the optimal GRNN forecasting model for one-step-ahead out-of-sample predictions. The results of the empirical study disclosed that the discrete monetary fundamentals are more informative in the Chinese and Malaysian exchange rates forecasting as compared to the relative monetary fundamentals. Nevertheless, seeing that the overall forecasting performance of the GRNN forecasting models underperformed the random walk benchmark model, the findings revealed that both discrete and relative monetary fundamentals failed to explain the dynamics of Chinese and Malaysian exchange rates except for the case of Chinese Renminbi vis-à-vis the US Dollar in the 12-month forecast horizon(This abstract was borrowed from another version of this item.)