This study analyses the prediction performance of the Asian Currency Unit (ACU) by employing variant methods including the Back-Propagation Neural Networks (BPN), Recurrent Neural Network (RNN), Time-Delay Recurrent Neural Network (TDRNN), General Autoregressive Conditional Heteroscedasticity (GARCH), and random walk models.
The results show that Artificial Neural Network models outperform GARCH and random walk models. The BPN model presents prominent forecasting performance in most division conditions. The study further verifies the causes of contagion of the Asian financial crisis using the Adaptive Network-Based Fuzzy Inference System (ANFIS). The empirical results indicate that the contagion effect would most likely be influenced by tight financial linkage and conditions of macroeconomic similarity as well.