Banking stability multifactor modelling in Armenia using Machine Learning

Abstract

Banking stability is a sensitive topic in economic literature and a lot of economists are trying to suggest better and better solutions, to foresee banking crises, and respond in a timely manner. Recent advancements in machine learning models as well as the increase in their usability, makes it inevitable their application in banking stability literature, especially when the policy makers are interested in early warning strategy and want to mitigate cumulative or systematic risks. The main concern about using ML models in banking stability tends to be the “Black box” side of neural network models, but this is compensated with their incredible predictive power, if used in a reasonable manner, taking into account best practices in a field. Apart from that, ML is not just limited to NN-s, Random forest approach suggests a way to understand which factors are more useful in the result of prediction.