I am working on a simple model with migration. The stochastic simulation works fine but for the non-linear model, the deterministic simulation does not converge even with a temporary shock to productivity.
To work around this issue, I log-linearized model by hand and used that for the deterministic simulation and it works fine for temporary shocks. I would prefer to get the non-linear model working because I want to look at transition dynamics after a permanent change in some model parameters which is less convenient to do (if I understand correctly) w/ a log-linearized model since it is linearized around the initial ss.
Additional note: In the investment decision, I am aware of the dynare timing convention, and the investment decision is purposely written differently here which is most closely related to Mandelman and Zlate (2012): ideas.repec.org/a/eee/moneco/v5 … 6-213.html
Please find .mod files for both attached.