Some methodological comments on 'Sources of fluctuations in exchange rates: a structural VAR analysis' by N. Apergis and C. Karfakis

In this comment we note that the methodology used by Apergis and Karfakis (1996) (a) leads to restricting the contemporaneous impact matrix not the long-run impact matrix in identifying their SVAR model, contrary to to their stated objective, and (b) that the Bernanke-Sims procedure of identification and the conventional Cholesky-decomposition based analysis of a VAR lead exactly to the same results as long as the impact matrix is restricted to be lower triangle.