Markov-Switching Models of Business Cycle: Can the Econometric Model detect the growth regime?

This article proposes first a univariate Markov-Switching Model of US GNP, decomposed into unobservable permanent and transi-tory components. Household consumption is then introduced into the model. The addition of this variable is interesting for two reasons:- test of the permanent income theory through a direct approach.- better detection of some recessions, notably those related to a deficit in global demand. We will see that this type of modeling also enables one to find, in real time, the different recessions which affected the last thirty years. Following this, an RBC model with indivisible labour, in which oc-cur two types of shocks (permanent and transitory governed by a first order Markov process), will be directly confronted with its natural empirical counterpart: the MS-UC bivariate model. Does the esti-mation carried out on simulated pseudo-observations allow us to find the reversal periods of the theoretical data generating process with greater accuracy