Title: Synchronization of cycles in a data-rich environment
Authors: Cem Cakmakli - Koc University (Turkey) [presenting]
Richard Paap - Erasmus University Rotterdam (The Netherlands)
Abstract: A general framework is proposed where the synchronization of the cycles embedded in multiple leading and coincident variables are modeled jointly. We use dynamic factor structure together with Markov mixture models to estimate the coincident and leading economic factors. The novel feature of the model is that we can estimate the coincident and leading economic factor together with their lead/lag relationship in a time varying manner. We show that lead time of the leading economic factor changes drastically over time. While the lead time increases during recession periods, it drops considerably at the onset of expansions. Finally, we provide evidence on superior predictive capability of our model over existing popular procedures in predicting key US macroeconomic variables.