Dating business cycles : a methodological contribution with an application to the euro area

This paper proposes a dating algorithm based on an appropriately defined Markov chain that enforces alternation of peaks and troughs, and duration constraints concerning the phases and the full cycle. The algorithm, which implements Harding and Pagan's non-parametric dating methodology, allows to assess the uncertainty of the estimated turning points due to filtering and can be used to construct indices of business cycle diffusion, aiming at assessing how spread are cyclical movements throughout the economy. Its adaptation to the notion of a deviation cycle and the imposition of depth constraints are also discussed. We illustrate the algorithm with reference to the issue of dating the Euro area business cycle and analyzing its characteristics, both from the classical and the growth cycle perspectives.