In a recent article Canova et al. (2014) study the optimal choice of variables to use in the estimation of a simplified version of the Smets and Wouters (2007) model. In this comment I examine their conclusions by applying a different methodology to the same model. The results call into question...