Theoretical moment Nan/ identifying unit root

[killergabin]

Dear all,

I'm running a model, everything seems good but I receive the following message "all endogenous are constant or non stationary not displaying correlations and autocorrelations" and I don't have theoretical moments. The purpose of my work is to obtain the historical decomposition of output, so i need help to identify the problem. Previous posts talk about unit root but I don't know how to identify the equation or the variables concerned.

Thank for your help.

Eric Gabin

[jpfeifer]

You have two unit roots in your model. It looks strange you have a variable P in your model. Usually prices in NK-models are non-stationary and everything has to be expressed in real terms using the inflation rate instead of the price level.

[killergabin]

You're right about that. I'm going to change that right now.

[killergabin]

Dear Johannes,
I tried to follow your advise but I'm not sure in the way to proceed. I replaced the price level in the model with an equation of exogenous inflation (even if I know it's not that simple) :
"PI(+1) = PIbar + RHOpi*(PI-PIbar) + ETA_pi". But now one of the eigenvalues is close to 0/0.

Can you give me additional advise?

[killergabin]

I also tried the New keynesian Philips curve " PI(+1) = RHOpi*PI + Kappa*Y"; but my Blanchard Kahn conditions are not satisfied.

[killergabin]

I think that I solve the unit root problem; but now theoretical moments have high or NaN std and variance. Thus I have bad variance decomposition.