Blanchard Kahn conditions

I know there are lots of posts regarding the Blanchard Kahn conditions, and I’ve read through lots of them. My problem is really confusing and non-representative though.

When I put in the original model, dynare gives me warning about Blanchard Kahn condition (Blanchard Kahn conditions are not satisfied: indeterminacy), but when I put in log-linearized model, dynare is able to simulate with no warning message.

I checked over and over again. I am sure there is no timing problem or any redundancy of equations.

Please help me out.

Attached are nonlinear model (every variable is log difference from its steady state) and linear model (every variable is log difference from its steady state) and the files of the steady state.

Thank you for your time
Sincerely
nonlinear_steadystate.m (3.1 KB)
nonlinear.mod (13.5 KB)
linearmodel_steadystate.m (2.82 KB)
linearmodel.mod (9.09 KB)

I run this code on version 4.1.2 is ok without bugs!