Une lecture probabiliste du cycle d’affaires américain

This paper explores 35 years of the American business cycle with the Hidden Markov Model (HMM) as a monitoring tool using monthly data. It exhibits ten US time series, which offer reliable information to detect recessions in real time. It also assesses the performances of different and complementary “recession models” based on Markovian processes : the “Pooled data model” and a multivariate HMM, and draws two main conclusions: simple HMM are decisive to monitor the business cycle providing that the series are proved highly reliable; models adding a multivariate dimension are useful but work marginally better than a simple summary : the inner quality of series seem to dominate their modeling. This paper introduces a new reading of the business cycle through, a favored recession model and concludes about leading and “real time detection” limitations. This paper is written in French.