A potential bug

[rockyyao]

Hi,

I use dynare to solve my model. I made a mistake on a coefficient in the model block, which is dependent on the steady state. I define all steady state value with x_bar. The mistake I made is that I forgot to write _bar, so I basically replaced the steady state variable with the log-deviation variable. (In this way, the model is no more linear). The strange thing is that Dynare still solves the model without an error message and reports some results. When I found the mistake and changed it. Dynare still solves the model and return some slightly different numerical results.

Can some one help me to figure out what is going on? Is it a bug or not?

I attach my model file with a function file needed for calculating the steady state. If you remove _bar from any n_bar in the first equation of the model block, you can see what I mean. I am using Dynare v 4.3.2.

Thanks

Rocky

[jpfeifer]

If you tell Dynare the model is linear by using

Code: Select all

model(linear);

Dynare takes you at your word and does not check whether your model is truly linear. This is not a bug, but a mistake on your part. However, we are currently in the process of adding a diagnostical tool to detect such cases. Moreover, Dynare 4.4 will check whether the steady state in linear models is satisfied.

[rockyyao]

Thanks for replying. I did define the model is linear by model (linear), nevertheless, Dynare solved the model with the mistake.

Ok, anyway thanks for your answer.