NaN AND inf in my steady state

[aadegboye abiodun]

Dear Dynare helpers,

Kindly provide a guide. I have NaN and Inf in three equations out of 22.

error no print. I attach the codes for you to read.

thank you

[jpfeifer]

The problem is equation 17

Code: Select all

1+(adebt/amp) = (adebt(-1)/api) + gove - atx;


amp and api are endogenous variables and consequently, not linear.

[aadegboye abiodun]

Dear jpfeifer,
I appreciate your effort and time. just within half a day i got response to my question.

The suggestion worked. wIth one or two modifications the model ran. However the simulation result seems less meaningful. DO you suggest I change my parameter values ?
your guide and help is required. kindly provide suggestions.

[jpfeifer]

Sorry, but I don't know your model and your problem with the results. Parameters should be chosen to match some target (or estimated). If these targets are off, change the parameters accordingly. If the targets are not off, but there is something else strange, check the model for correctness or, if it is correct, modify the model.

[aadegboye abiodun]

Dear jpfeifer,

once again I appreciate youe effort and time. i herewith my model codes. just forme to be very certain, could you kindly mention one or two of the targets you mentioned that the parameters must agree with. for the results some graph could not be clearly identified while some appear not to be seen.
your guide would be helpful.

[jpfeifer]

For example, the discount factor is usually set to be consistent with a long-term interest rate. The openness parameter is sometimes set to a values consistent with the openness of the economy and so on. For autocorrelations and shock standard deviations, the target is often something like output and investment volatility.

[aadegboye abiodun]

dear jpfeifer,
thank you for the effort and time once again my model can now run. I checked the correctness of my model and so far I have not figure out any mistake. However I have NaN in some of my theoretical moments of the endo variables while others have zeros in almost all. Also if the impulse-response graph does not return to steady state does it mean there are issues someone? if yes your sugestions would be welcome and helpful.

Once I ensure my model is okay I want to immediately take it data.

yours

[jpfeifer]

All of this is explained by the presence of a unit root. In this case your IRFs will have permanent effects and the unconditional second moments for the non-stationary variables do not exist, i.e. are NaN. Check whether the unit root is a feature of your model. Search the forum for more on this. There is no universal solution.