Economic policy uncertainty and bankruptcy filings

Applying machine learning techniques to predict bankruptcy in the sample of French, Italian, Russian and Spanish firms, study demonstrates that inclusion economic policy uncertainty (EPU) indicator into prediction models notably increases their accuracy. This effect is more pronounced when we use novel Twitter-based version EPU index instead original news-based index. We further compare accuracy conclude stacking ensemble method outperforms (though marginally) methods, which are commonly used for prediction, such as single classifiers bagging.