Complex open economy steady state

Dear all,
I am working to extend a medium large scale model in to an open economy. The framework for open economy is from Gali and Monacelli and the model is from Chang et al (2019) (https://doi.org/10.1016/j.jmoneco.2018.08.005) I know that the model is quite complicated and i tried my best to get the analytical steady state. However, i cannot see how to solve it anymore. Any advice is extremely appreciated because i hit the wall. Thank you so much for all the helps.calibration1.m (1.7 KB) console_nkopen.m (7.2 KB) csss.m (3.3 KB) dsss.m (3.7 KB) gdpsss.m (3.7 KB) nk_open.mod (14.5 KB) par.mat (2.8 KB) shocktype.mat (176 Bytes) solveff.m (1.4 KB)

Please solve the models in levels first, i.e. without the logs. Your tb is negative, explaining the complex values when you take the log.

On Zheng Liu’s website

You will get the supplemental appendix and the replication code.

Hi,
Thank you a lot for the comment. It works somehow now by solving non linear model. However, i have not came out with analytical solution yet but have to use the Dynare steady solver. Can it be a problem that the solver gives a wrong not suitable steady state ? I will try more but i am just curious.

There are two issues.

  1. If there are still mistakes in your model, an analytical steady state will help finding them. See e.g.


2. In case your model has multiple steady states, the solver may converge to the wrong one.

As a minimum, you should check whether the steady state makes sense.

Hi, sorry for more question. I obtained some IRFs but it seems weird and two eigen values are complex. I looked around and this may be problem of timing issue. Can it be ? Or my model is just wrong in structure. I am new to this and so sorry for those questions. Thank you so much.

That’s hard to tell. Oscillating IRFs like that are usually indeed a sign of timing error. But it can also be a misplaced sign error or something like that.
Or course your model structure could be wrong, but then it would be rather unlikely to run at all.