Identifiability of Dynamic Stochastic General Equilibrium Models with Covariance Restrictions

In this paper, we study the identification problem of parameters of Dynamic Stochastic General Equilibrium Models with emphasis on structural constraints, in order to make the number of observable variables is equal to the number of exogenous variables. We derive a set of identifiability conditions and suggest a procedure for a thorough analysis of identification at each point in the parameters space. The procedure can be applied, before DSGE models are estimated, to determine where identification fails. We also use a Monte Carlo simulation and study the effect of restrictions on the estimate. The results show that the use of restrictions for estimation, when identification is reduced, leads us to inaccurate estimates and unreliable inference even when the number of observations is large.