Economic Recession Prediction using Machine Learning: A MIDAS Approach

This article analyzes the prediction of the recession in the United States by combining machine learning (ML) algorithms and a mixed data sampling approach (MIDAS) to construct the MIDAS-ML models. Some of the most representative machine learning techniques are implemented for modeling and forecasting U.S. recessions. The empirical analysis shows that the MIDASML models outperform the benchmark MIDASLogit/Probit model in prediction abilities, as indicated by the evaluation of various statistical metrics.