Hi experts!

I want to know if ramsey_policy has an option to do second order approximation.

Thanks for your help

Hi experts!

I want to know if ramsey_policy has an option to do second order approximation.

Thanks for your help

Hi,

I’m not really an expert, but in standard dynare you can only use order=1, as far as I know. You can use Dynare++ to get higher-order approximations, though.

Furthermore, there is a possible workaround if you want to use standard dynare (via dynare++) to simulate the model - the reason for doing this is that I think there are more options in dynare than in dynare++. If you run your file in dynare++ asking it to do the ramsey policy, dynare++ produces a dmp file containing the model, not only the model restrictions (equations (1) in the link below), but also the first-order condition with respect to the economic variables (equation (3) in the link). I think the resulting system describes the economy under the Ramsey policy. Then you can use the contents of the dmp file to create a mod file and then just use stoch_simul with the appropriate order. An advantage compared to dynare++ is that you e.g. can use different solutions algorithms instead of the dynare++ algorithm.

Does anyone in the Dynare team know if the workaround is ok?

Best

Andreas

Standard Dynare goes up to order 3 with stoch_simul.

But for Ramsey policy you are limited to order 1.

Thanks!!!

Hi,

I have used the approach of Schmitt-Grohe and Uribe that Andreas had pointed out to. I have checked the derived equations many times and I believe they are correct - I cannot spot a mistake, BUT I get completely ridiculous IRFs (attached).

Any suggestions why I could get such weird IRFs & persistence??? Btw, I have tried the approach by Schmitt-Grohe and Uribe using a smaller-scale model and it works perfectly fine.

Full page photo.pdf (42.6 KB)

[quote=“SébastienVillemot”]Standard Dynare goes up to order 3 with stoch_simul.

But for Ramsey policy you are limited to order 1.[/quote]

Hi Sébastien,

What is the reason for limiting Ramsey policy solutions to order 1? I’m using Andrew Levin’s get_ramsey package, and have noticed that we can solve the .mod file with Ramsey equations fine in first-order but when we use second-order we are unable to get the theoretical mean for many variables (from oo_.mean). The same file with an instrument rule instead of the Ramsey equations returns the theoretical mean when calculated to second-order. I’m wondering if this is something inherent in calculating Ramsey policy or something wrong with our .mod file.

Thanks,

Joe

I think it is just that nobody had the time to implement higher orders.