Estimate my own state space model

Estimation ML/Bayesian estimation
1

I used Dynare to construct a rational expectation DSGE model and retrieved the first-order approximation of the model’s equilibrium conditions and the state-space representation of the solution through dynamic_g1.m and oo_.dr. Based on this benchmark model, I incorporated a bounded rational expectations formation mechanism. After doing this, the model’s actual law of motion (ALM) can still be written as a linear state-space model, but its specific form differs from that of the original rational expectations model, and some endogenous variables that were originally observables have now become state variables.

Since I know that writing my own estimation algorithm might lead to many errors, I would like to know whether it is possible to input my modified model into one of the intermediate steps based on Dynare’s Bayesian estimation, and then use Dynare’s built-in algorithms to estimate my model.

Thank you in advance.

2

Do you now have a new representation of the model as a function of the parameters? Or would you need to solve the RE model first for each set of parameters, do the transformation, and then obtain the new state space representation?

3

Thanks for your reply! Now I need to solve the RE model first by Dynare for each set of parameters. But if the representation of the model as a function of the parameters must be computed by myself, I believe I can manage to do that.