Hi everyone,
I have a question about ramsey_policy in dynare version 4 and get_ramsey code of Levin. Please help me.
I am working on model with borrowing constraints. I tried to get_ramsey to solve for the optimal monetary. For the last two months I tried to work get_ramsey but all I get is BlanchardKahn condition. The dynare output tells
There are 13 eigenvalue(s) larger than 1 in modulus
for 12 forwardlooking variable(s)
The rank condition ISN’T verified!
One of the eigen values is equal to 1.000265427 and it is counted as greater >1. Is there a way to tell which variable in my model is causing this problem or which one of the equations?
I also tried to ramsey_policy with dynare 4. Although in this case I am able to get IRFs , the model is very sensitive to the initial value of the lagrange multiplier in front of the borrowing constraint in the model.
The model works perfectly fine for simple interest rate rules but gives me a very hard time when I try to get optimal monetary policy either with ramsey_policy or get_ramsey.
My questions are

is there way to tell by looking at eigen values where the problem comes from?
I modified the model, I changed the parameter values. I am positive that the model’s steady is correct. The model with a simple Taylortype rule converges to the steady state I found. I tried millions things, but I was not successful in figuring out to solve the BK problem. 
How does ramsey_ramsey policy differ from get_ramsey?
I know with ramsey_policy, we do a linear approximation to the model and we define a sort of loss function. But other than these, I really cannot understand why my model work with ramsey_policy (although it is very sensitive to langrange multiplier in front of borrowing constraint) but not with get_ramsey.
Thanks in advance.
Best,