A0276
Title: Trend in Markov-switching VAR models
Authors: Maddalena Cavicchioli - University of Modena and Reggio Emilia (Italy) [presenting]
Abstract: The purpose is to consider Markov-switching vector autoregressive (MS VAR) processes that incorporate both a stationary component and a deterministic time trend. These models are particularly relevant for capturing structural changes and asymmetries in macroeconomic time series. To estimate the model parameters, the ordinary least squares (OLS) method is adopted within a modified expectation-maximization (EM) algorithm framework. This approach offers computational simplicity while accommodating regime changes governed by an unobserved Markov process. The consistency is established, and the asymptotic distribution of the resulting OLS estimators is derived. To the best of knowledge, the characterization of the asymptotic variance-covariance matrix for OLS estimators in this context fills a gap in the existing econometric literature. An empirical application focused on housing market asymmetries demonstrates the practical relevance and effectiveness of the proposed methodology.
