error in bayesian estimation, help please.

[lssp]

Hi, Professor Pfeifer,

When I was doing the bayesian estimation, I encountered the problem like this:

Warning: Matrix is close to singular or badly scaled. Results may be
inaccurate. RCOND = 5.142215e-17.

I am quite sure that my model is right, and then what may the problem be ? and how can I solve this ?

Any advice will be appreciated.

Thank you so much!

[jpfeifer]

For starters, you are estimating too many parameters given your data. You have only one observed data series and this makes some of the parameters not identified. Either add additional observables or estimate fewer parameters. To see this, put

Code: Select all

identification;

instead of estimation (Important: do not use estimation after identification; use either the one or the other to avoid unwanted interaction between the commands)

[lssp]

Hi, Professor Pfeifer,

Thank you for your reply, I really appreciate that.

Now I have another question about the observation equation after I red your paper "A Guide to Specifying Observation Equations for the Estimation of DSGE Models".

Q:
The model I am dealing with (as attached already in the last post) is a nonlinear model, but the Taylor Rule is linear form, like:

ReXU = rhotilUU*ReXU(-1) + (1-rhotilUU)*(ReXUU+aptilUU*(piU(+1)-piUU)+aytilUU*(YU-YUU)) + e_xpU;

If I want to do the estimation using the data of ReXU, how should I write the correct observation equation ? and how to deal with the raw data ?

I am looking forward to your reply, thanks again!

[jpfeifer]

See http://www.dynare.org/phpBB3/viewtopic.php?f=1&t=6318