摘要：We introduce a new approach for the estimation of high-dimensional factor models withregime-switching factor loadings by extending the linear three-pass regression fi lter tosettings where parameters can vary according to Markov processes. The new method,denoted as Markov-switching three-pass regression fi lter (MS-3PRF), is suitable for data setswith large cross-sectional dimensions, since estimation and inference are straightforward,as opposed to existing regime-switching factor models where computational complexitylimits applicability to few variables. In a Monte Carlo experiment, we study the fi nite sampleproperties of the MS-3PRF and fi nd that it performs favourably compared with alternativemodelling approaches whenever there is structural instability in factor loadings. For empiricalapplications, we consider forecasting economic activity and bilateral exchange rates, fi ndingthat the MS-3PRF approach is competitive in both cases.
