# Question about theoretical moment after implementing ramsey_policy

Dear jpfeifer

Hello.

1.Can I ask you how is theoretical momet calculated after implementing ramsey_policy?

Are they related to the moment of IRF?

1. In addition, is it problematic if there exists a unit root? Does that mean my model is wrong or need correction such as stationarization? I think that Dynare can not calcaulate the theoretic moment for some variables because of the unit root.

I attach my mod and data file.
lumpsum_noSharing_Ramsey_nw.mod (6.9 KB)
lump_noS_params.mat (653 Bytes)

1. Where can I find the explanation about opt_algo (in osr procedure)?
I used opt_algo=1, is it same with the Newton-type nonlinear solver?

Thank you in advance

Best regards
Seungcheol

1. After setting up the first order conditions of the planners problem, Dynare conducts a first order approximation. This results in a linear state-space system representing the solution. See

Yes, they are related to IRFs in the sense that in each period shocks realize that trigger IRFs. The sum of the respective IRFs at each point in time causes the variables to move. The moments are then computed on these variables (although only conceptually if you consider theoretical moments)

1. Your code loads a `lump_noS_params_nw.mat`, but you only provided a `lump_noS_params.mat`. Unit roots are not a problem, if they come from economic features of the model as opposed to coding errors. Of course, the unconditional second moments of the unit root variables do not exist as they are infinite.

2. The Dynare manual describes which optimizer is used. `opt_algo=1` triggers Matlab’s `fmincon`. See the section “Algorithm” at Find minimum of constrained nonlinear multivariable function - MATLAB fmincon - MathWorks Deutschland

Thank you jpfeifer. Your explanation is really helpful!

1. So, can I understand that the theoretical moment is based on the IRF for random shocks series (simulation)?

2. I upload correct data file again. Can you just run my file?
lumpsum_noSharing_Ramsey_nw.mod (6.9 KB)
lump_noS_params_nw.mat (635 Bytes)

Thank you again!!

Best regards
Seungcheol

1. No. As I said
1. I ran your code. The unit roots seem strange.

Dear jpfeifer

Thank you so much for your quick and helpful reply.

Can I ask you why the unit roots seem strange?

Furthermore, is there any way to fix this?

When I use model_diagnostics, it says, for example,

Colinear equations
11 20
even though the number of my equations is 17.

In this case, what is the 20th equation?

Thank you in advance.

Best regards
Seungcheol

1. Because it affects `cco` and `ccr` and unit roots in consumption are uncommon. But total consumption `CC` is not affected. This indicates that shocks permanently shift consumption from one type of agent to the others. Maybe this is a feature of the model.
2. You are solving a Ramsey problem. Thus, there are equations of the Ramsey problem added to the problem. These are the first `M_.ramsey_eq_nbr` equations, followed by the `M_.orig_eq_nbr` original equations, resulting in `M_.eq_nbr` total equations.

Dear jpfeifer

Thank you so much!

Best regards
Seungcheol