Can I estimate the model, which I have solved?

Hi,

I just began my learning of dynare. I solve a three-equation New Keynesian Model (Phillips, IS and Taylor rule) with zero lower bound, using the method provided by www2.bc.edu/matteo-iacoviello/r … E_2015.pdf

I attach the code. You can solve the model by run “runsim_cgg.m”.

However, I don’t know how to estimate this model…Can I estimate this based on Iacoviello’s method? Can I estimate in dynare??

It would be appreciated if anyone can help me! Thanks a lot!
model_cgg.zip (42.5 KB)

You cannot use likelihood based techniques as Dynare cannot construct the likelihood.
You could try to do moment matching for example, but I am not sure this is possible with OCCBIN. I still have to read the paper.

[quote=“jpfeifer”]You cannot use likelihood based techniques as Dynare cannot construct the likelihood.
You could try to do moment matching for example, but I am not sure this is possible with OCCBIN. I still have to read the paper.[/quote]

Thanks very much for your help!

Another questions, I don’t know why he write the Taylor rule of interest rate (equation 7.1 in page 35 in the attachment) as

“rnot = FIRrnot(-1) + (1-FIR)(FIPp + FIYy)” in cgg.mod and cgg_zlb.mod

May you help me about this? Thanks very much!
science.pdf (555 KB)

[quote=“jpfeifer”]You cannot use likelihood based techniques as Dynare cannot construct the likelihood.
You could try to do moment matching for example, but I am not sure this is possible with OCCBIN. I still have to read the paper.[/quote]

Besides, why do they only consider the demand shock in cgg.mod? I think we can’t ignore cost-push shock and monetary policy shock in the three-equations NK model. Also I think we can consider the three shocks at the same time…I think we can just add another two shocks, named ‘eps_u’,‘eps_m’ as ‘eps_g’ and add u and m into the linear equations. Then give them initial values etc.

OCCBIN is just for simulation, do estimation ,you need to write code yourself.

Hi! I also have a question how to estimate with Bayesian methods a model solved with Occbin?
Is it like you have to estimate first parameters for both regimes separately or how should if function?

You have to look at Iacoviello’s new paper. The basic Occbin was only for simulation, not estimation.