Kalman Filter Aguiar Gopinath (2007)


I want to replicate the model by Aguiar and Gopinath in an imperfect information setting. Representative agent will be imperfectly informed about the decomposition of TFP shocks into its trend growth and transitory components and will form expectations of the decomposition using Kalman Filter. I wrote the state space representation (mathematically, on paper) following Harvey (1989). Can I do this in Dynare? If I can, how should I do it? I am very new to Dynare.

Thank you very much.

If you have a state space model on paper, I think dynare can estimate that if some conditions are met. For example, if the number of state equations matches the number of state variables and BK conditions are satisfied. The DSGE model is already a state-space model, yeah? Or maybe your state space representation is a little different from what is typically seen in DSGE dynare examples?

This can be done in Dynare, but is a bit more sophisticated. An example is Hürtgen (2014). You essentially have to solve the Kalman filter problem manually in a steady state file.

I cannot understand what you are saying. I expressed the -filtering problem- I have in mind in state space form (I forgot to mention but I will give one additional noisy signal other than TFP shock). I mean, I formed my measurement equation, transition equation and so on following Harvey. I just have to translate it into Dynare. What I am asking is, how can I combine the filtering code with the model code?

Thank you very much professor.