Analysis Of Financial Instability By Means Of Decision Trees And Lists

This paper tries to further investigate the factors behind a financial crisis using a large sample of countries in the period 1981 to 1999.Most methods applied in other researches to analyse financial instability are techniques of statistical nature and the variables employed in these models do not usually satisfy statistical assumptions, what complicates the analysis.In order to avoid these inconveniences, we have approached this problem by applying two algorithms coming from the Machine Learning field, the C4.5 and PART algorithms. By means of theses algorithms we will try to analyse the role of a set of macroeconomic and financial variables, both quantitative and qualitative, in explaining banking crises.Results indicate a high performance of the machine learning techniques, what shows that these methods can be a useful tool to evaluate financial instability and a competitive alternative or complement to existing models related with this kind of problems.