Predicting the Probability of Financial Crisis in Emerging Countries Using an Early Warning System: Artificial Neural Network

Abstract

The first upsurge of rising economies in the world took place in early 2000s in Brazil, Russia, India, and China, which are known as BRIC countries. Since then, these countries have demonstrated great investment opportunities for financial services as well as the other real sectors’ services. Last decades have witnessed frequent financial disruptions in different kinds which resulted in ruining the economy with unwanted social consequences in BRIC as well as the other countries. The destructive consequences of crisis explain the major reasons for estimating the predicted probability of the crisis. This paper studies the factors associated with the emergence of financial crisis in BRIC countries during 1992-2011 using an Artificial Neural Network. In this context, we built, trained, and tested an Early Warning System (EWS) in order to find out the importance of different inputs in explaining the crisis. Socalled model has proven itself by predicting the crisis and non-crisis dates very well. Comparing the importance and significance of all variables in the model, it was discovered that the domestic credit to private sector (% of GDP, domestic credit growth), Inflation, freedom, interest rate and economic growth were the most significant variables in this model, while the deposit insurance rate was found to be the least significant variable in explaining the crisis.