A multivariate system for turning point detection in the euro area : 2016 edition

In the paper we introduce a system for statistical detection of turning points in the euro area based on the class of multivariate Markov-switching models. Component series considered in the application are monthly short term statistics released by Eurostat, business and consumer surveys released by DG-ECFIN and new cars registrations released by ACEA. The main advantage of the proposed approach is that it jointly derives a couple of coincident indicators for both the classical business cycle and the growth cycle; further, it assures the sequence of turning points to be coherent with the ABCD approach, an occurrence which is not guaranteed by applying the univariate approach for the pair of coincident indicators. The application considers an extensive combination of multivariate setups, comparing the results with those related to the univariate approach. Selection of the most appropriate multivariate system is carried out by adopting a set of statistical benchmarking criteria with respect to a reference dating chronology.