The official institutions (NBER, OECD, CEPR and others) provide business cycle chronology with a lag from 3 months up to several years. Markov-Switching Dynamic Factor Model (MS-DFM) allows to produce the turning points more timely. The Kalman filter estimates of the model can be obtained in one step with limited number of series or in two steps on a much richer dataset. While the choice of correct series is a challenge for the one-step method, the problem of the two-step method is the potential misspecification. In this paper we apply one-step and two-step approaches to the French data and compare their performance. Both methods give qualitatively similar results and prove to reproduce the OECSD business cycle chronology on the 1993-2014 monthly sample well. We find that the two-step method is more precise in determining the beginnings and the ends of recessions. Also, both methods produce extra signals corresponding to downturns which were too short to belong to OECD chronology of recessions.