We propose a classical approach to estimate factor-augmented vector autoregressive (FAVAR) models with time variation in the factor loadings, in the factor dynamics, and in the variance-covariance matrix of innovations. When the time-varying FAVAR is estimated using a large quarterly dataset of...