CMStatistics 2021: Start Registration
View Submission - CFE
Title: Dynamic mixture vector autoregressions with score-driven weights Authors:  Dennis Umlandt - University of Trier (Germany) [presenting]
Alexander Georges Gretener - University of Kiel (Germany)
Matthias Neuenkirch - University of Trier and CESifo (Germany)
Abstract: A novel dynamic mixture vector autoregressive (VAR) model is proposed in which time-varying mixture weights are driven by the predictive likelihood score. Intuitively, the state weight of the $k$-th component VAR model in the subsequent period is increased if the current observation is more likely to be drawn from this particular state. The model is not limited to a specific distributional assumption and allows for straightforward likelihood-based estimation and inference. We conduct a Monte Carlo study and find that the score-driven mixture VAR model is able to adequately filter the mixture dynamics from a variety of different data generating processes which most other observation-driven dynamic mixture VAR models cannot cope with appropriately. Finally, we illustrate our approach by an application where we model the conditional joint distribution of economic and financial conditions.