Matteo Barigozzi on Dynamic Factor Models, Cointegration, and Error Correction Mechanisms

Speaker:  Matteo Barigozzi - London School of Economics
  Monday, March 24, 2014 at 12:30 PM Aula H

In this paper we study Dynamic Factor Models when the factors Ft are I(1) and singular, i.e. rank(Ft) < dim(Ft). By combining the classic Granger Representation Theorem with recent results by Anderson and Deistler on singular stochastic vectors, we prove that, for generic values of the parameters, Ft has an Error Correction representation with two unusual features: (i) the autoregressive matrix polynomial is finite, (ii) the number of error-terms is equal to the number of transitory shocks plus the difference between the dimension and the rank of Ft. This result is the basis for the correct specification of an autoregressive model for Ft. Estimation of impulse-response functions is also discussed. Results of an empirical analysis on a US quarterly database support the use of our model.

Title Format  (Language, Size, Publication date)
Paper  pdfpdf (it, 821 KB, 19/03/14)

Programme Director
Andrea Roventini

External reference
Luigi Grossi

Publication date
November 30, 2013