Ireland (2004) example dynare file


#1

Hi everyone,

I’m looking at the Ireland 2004 replication mod file from github.com/johannespfeifer/dsge_mod

It seems Johannes has the file set to simulation with the estimation commented out. If I comment out the stoch_simul and comment the estimation procedure back in, I can’t seem to get it to run. Specifically with errors in computing likelihood for initial parameter values. Can anyone get the estimation in this example to work?

Thanks,
J


#2

That’s strange. I can run the example with estimation. What code exactly are you using? And which Dynare version?


#3

I’m running the file attached here, where I’ve just deleted the plots at the end, commented out ‘stoch_simul’ and commented back in the estimation portion.

Using Dynare 4.4.3, I get the error quoted below.

Also, while I’ve got your attention, I think the current ‘Ireland_2004.mod’ file on your Github has a typo on line 172 in the estimated params block (omega isn’t specified and I couldn’t run it as is)

Ireland_2004_est.mod (6.29 KB)


#4

The mod-file is fine as is, but you need to use the unstable version/4.5. See the disclaimer at github.com/JohannesPfeifer/DSGE_mod


#5

Hello,

I’m having a similar problem with my own code. When trying to estimate my model with mode_compute=9, I’m getting the following warning

STEADY: numerical initial values or parameters incompatible with the following equations 12 17

However, just before estimation when using
resid(1);
everything seems ok, so I do not know how can I get this warning…

I’m using Dynare 4.5.3 in mac.

Regards,

Veronica

CESEstFinal.mod (7.4 KB)


#6

The dataset is missing for running the code


#7

Sorry. Here you have it.estimation.xls (22 KB)

The code runs perfectly with mode_compute=5 or 6.


#8

Dear Johannes,
Sorry to disturb you but do you have any idea of what is happening with the error? I just upload the estimation file so you can run it.

Regards,
Veronica


#9

The reason is that mode_compute=9 tries a wider range of parameter values and apparently encounters some parameter draws that result in invalid expressions, i.e. NaN and Inf. But as long as it’s only a warning I would not worry.


#10

Perfect, thanks a lot!