Estimation of DSGE Models (Bayesian Econometrics)


I have one small specific question regarding the estimation of parameters of a DSGE model hereby I have 8 final log-linearised equations. Do I need a dataset for each equation, so in this case 8?

The Problem here is that three endogenous variables are among others the shadow prices of the constraints (2 Lagrangian multipliers) or real marginal costs and are therefore not observable or available on a quarterly basis.

Could I solve this problem by reducing the equation system by substituting the final equations in each other?

Thank you very much!

Since you have 8 equations, you should also have 8 endogenous variables. You don’t need a dataset that contains all of the 8 variables. Just gather data on some of the variables that are observed and link these observed variables to the model variables by specifying observation equations.

1 Like