The estimation of factors in FAVAR models

Abstract: The use of factor-augmented vector autoregression (FAVAR) models has become increasingly popular in the literature of empirical macroeconomics. This paper sheds light on the different factor estimation methods that are available to researchers. More specifically, this paper examines the widely used principal component method but also the computationally simpler common correlated effects method as well as the more advanced likelihood-based method using the Gibbs sampler. The results indicate very little difference between the principal component method and the common correlated effects method, which can facilitate the estimation of FAVAR models for researchers within the field of macroeconomics.