This. paper analyzes identificat.io~ conditions, and proposes an estimator, for a dy-namJc factor model.where the 1d10syncratic components are allowed to be mutually non-ort~ogonal. ThJS mode!, which we cali generalized dynamic fador model is nove! to. the hterature,. an d generalizes the stati c approximate • factor mode! of Chamberlam and Rothschild (1983), as well as the exact factor mode! à la Sargent and Sims (197~). We prove rr:ean-s~uare convergence of our estimator to the common factor as t~e t1me cross-~ect10nal d1~ensions go to infinity at appropriate rates. Simulations yJe!d encouragmg results m small samples. An empirica! example on the out t growth of US states illustrates the method. pu JEL classification nos.: C13, C33, C43.

SUMMARYThis paper shows that vector auto regression (VAR) with Bayesian shrinkage is an appropriate tool for large dynamic models. We build on the results of De Mol and co-workers (2008) and show that, when the degree of shrinkage is set in relation to the cross-sectional dimension, the forecasting performance of small monetary VARs can be improved by adding additional macroeconomic variables and sectoral information. In addition, we show that large VARs with shrinkage produce credible impulse responses and are suitable for structural analysis.

No abstract

4Non-technical summary 5