Hi,

I would like to estimate my model recursively (adding one additional period each time) over the sample and I am using the nobs = [initial period:final period] option to do it, which works very well.

For each additional period, Dynare computes the model’s mode, starting from the one found at the previous iteration.

What I would like to do, is to compute the mode from zero at each iteration. I have tried with the optim=(‘SaveFiles’,0) option but it still starts from the last mode.

Is there any way?

Thank you in advance!

Answering to my self, a possible solution would be to loop over different bayesian estimations.

Is there a way to do it? It looks like dynare doesn’t allow to set something like:

for i=90:100

estimation…

nobs=i

…

end

Otherwise I would write a simple code in matlab calling the (different) mod files.

Thanks!

What do you mean with

?

Dynare usually starts the mode-finding from the prior mean unless a mode-file is specified.

Yes, sorry, I meant from the prior-mean.

When doing a recursive estimation (adding one period at a time), the mode-finding starts at the prior-mean for the first iteration. Each following iteration, however, starts from the last posterior-mode.

Indeed, at each iteration Dynare saves a new mode-file, which is then used as starting value for the next mode-finding. This makes sense, but my goal would be to run a recursive estimation starting each iteration at the prior-mean.