hi everbody,

does anyone have an idea how to change Dynare to be able to solve models where the Blanchard-Kahn conditions are not satisfied?? the code always terminates in case of indeterminacy.

I’m grateful for all answers

saludos, chris

I’m really not an expert in this topic, but have a limited experience that can be useful:

I don’t think you should make changes in Dynare (I mean in the .m code files) unless you are absolutely sure of what you are doing. When considering rewriting the control file (.mod), this you should know:

BK can fail on at least two grounds:

first and foremost you should CHECK THE TIMING of your variables very carefully (this solved 90% of the cases when I received this message). In general you should check your whole theoretical model for stability.

second: it can be a numerical problem (e.g. you log linearize a concave production function at a point where the returns to scale is high (>>1)). This can be amended if you try to choose parameters so that the steady state of your model be free of this kind of instability.

I am not very sure about what kinda of result you expect from the model with indeterminacy. It simply means everything is possible, so the IRFs are won’t converge. The literature on indeterminancy usually focus on when indeterminacy gonna happen. With Dynare, you surely can do the job by collecting the parameter values that falls in the indeterminant region.

On second thought: if you really want to use a theoretical model whith multiple equlibria then technically you can solve for one of these equilibria for example by pinning down as a parameter the initial value of an appropriate variable and model thereafter the changes of that variable.

E.g. in a Calvo pricing model, price level can be arbitrary, but you must model the changes in it, that is inflation.

On third thought: could you clarify what kind of indeterminancy you would like to model (and why)?