Real-time Recession Probability with Hidden Markov Model and Unemployment Momentum

Abstract

We show how to construct a composite Hidden Markov Model (HMM) to calculate real-time recession probability, using the jubilee and ldhmm packages in R. The input data is the unemployment rate (UNRATE) which is released monthly by the U.S. government. There are two sub-models: The one-year momentum model and the 6-month acceleration model. The product of the two generates the recession probability. Our model demonstrates that positive momentum in unemployment kicks off a recession. The momentum accelerates during the recession. And eventually the rapid deceleration marks the end of it.