On the validity of exclusion restrictions in the structural multivariate framework: a Monte Carlo simulation

摘要

This paper aims to examine the validity of identifying restrictions used in the structural multivariate models. Whether we are under short-term and/or long-term identification approach, additional restrictions must be imposed and usually take the form of exclusion restrictions. We believe, however, that the value of a restriction is not necessarily equal to zero even if it expresses the lack of impact of a shock on a variable. Such a lack of impact could reflect an effect asymptotically equal to zero and the little nuance could be amplified with the model dynamics. To do, a Monte Carlo simulation is performed to examine the consequences of slipping of the identification restriction value. The results confirm the sensitivity of variables' responses to change in the value of identification restrictions. Whatever the strength and elegance of the theory and the economic reasoning from which emanate the exclusion restrictions, precision measurements should be considered.