SVARs Identification Through Bounds on the Forecast Error Variance

This article identifies structural vector autoregressions (SVARs) through bound restrictions on the forecast error variance decomposition (FEVD). First, shows FEVD bounds correspond to quadratic inequality columns of rotation matrix transforming reduced-form residuals into shocks. Second, establishes theoretical conditions such that lead a reduction in width impulse response identified set relative only imposing sign restrictions. Third, this proposes robust Bayesian approach inference. Fourth, elicitation could be based DSGE models with alternative parameterizations. Finally, an empirical application illustrates potential usefulness for obtaining informative inference set-identified monetary SVARs and remove unreasonable implications