Bayesian Estimation of DSGE models: Is the Workhorse Model Identified?

Koop, Pesaran and Smith (2011) suggest a simple diagnostic indicator for the Bayesian estimation of the parameters of a DSGE model. They show that, if a parameter is well identified, the precision of the posterior should improve as the (artificial) data size T increases, and the indicator checks the speed at which precision improves. It does not require any additional programming; a researcher just needs to generate artificial data and estimate the model with different T. Applying this to Smets and Wouters'(2007) medium size US model, we find that while exogenous shock processes are well identified, most of the parameters in the structural equations are not.