Fatemeh Farzin - Faculty of Managements and Economics, Tarbiat Modares University, Tehran, Iran
Ahmad Googerdchian - Faculty of Administrative and Economics, University of Isfahan, Isfahan, Iran
Babak Saffari - Faculty of Administrative and Economics, University of Isfahan, Isfahan, Iran

Today, economists have paid much attention to prediction of currency crises due to their very negative effects on the performance of real economies and the consequences of its subsequent recession. The use of an early warning system for currency crises is therefore introduced as an empirical tool for troubleshooting Iran macroeconomic problems. Based on studies conducted in other countries and using conventional methods of extracting symptoms and estimating crisis probability, an early warning system for currency crises is presented to the Iran economy that can warn currency crises beforehand. Using Multi-layer Perceptron Neural Network and Hard-Limit function, the Early Warning System is basically design to consider seasonal data for the period of ۲۰۰۱ to ۲۰۱۵ to anticipate currency crisis of ۲۰۱۹(based on ۱۲-season warning periods). The results predicted show that no currency crisis is threatening Iran economy in ۲۰۱۹. The export index is one of the leading variables in the system which has the greatest impact on currency crises. In addition, according to the previous data and their substitution in this model, the years ۱۹۹۳, ۲۰۰۱ and ۲۰۰۳ are signaled as critical years. It therefore can be concluded that the present research model is reliable.

early warning system, Currency Crisis, Artificial Neural Network, Perfect Signal