Chauvet (1998) developed the Smoothed U.S. Recession Probabilities, and it is published by The Federal Reserve of St. Louis. This paper points out the shortcomings of this probability index and proposes an alternative index using statistical learning methods. This alternative index of recession probabilities is built using Support Vector Machines over the same range of monthly data on coincident indicators. The alternative index obtained exhibits slight improvements compared to the current index; specifically, the alternative index provides more accurate probabilities over several intervals of time. |