Page 1 of 1

The rank condition ISN'T verified_1 colinear relationship

PostPosted: Sat May 24, 2014 1:31 pm
by nk178
Dear Dynare users,

It is the first time I run Dynare for the first chapter of my PhD thesis. I develop a two-country model for currency union with unemployment and I want to solve the model in order to observe the IRF's of a productivity shock in each country.

The result I get is:

'' There are 7 eigenvalues larger than 1 in modulus for 7 forward-looking variables, The rank condition ISN'T verified! ''

It seems that the necessary condition for uniqueness of a stable equilibrium is satisfied but probably is not a sufficient condition.

as it also says that '' Blanchard Kahn conditions are not satisfied:indeterminacy due to rank failure''.

I guess this means that the A matrix of the canonical form is not full rank so it is not invertible.

I run the diagnostics and say that there is 1 colinear relationship btw variables and equations, and then it says that 8 out of 12 variables are colinear!!

Notice that I have already log-linearized the equations of the model by hand (in order to observe the proxy of inflation in the New Keynesian Phillips curve) so actually in the model command I type linear and I set initval SS values equal to zero (as it supposed that all of them are logs of numbers smaller than 1). I don't know if this is relevant to the colinearity problem or not.

I am very confused as I don't know where the problem is so I don't know what to do, like:

1)Shall I change the value of the model parameters?
2)Shall I change the specifications of the equations of the model? What is suggested?
3)Shall I change the taylor rule?
4) Shall I log-linearize again? ( I need to do by hand to obtain the proxy of inflation endogenously)
5) Are the colinear relationships obtained because the model is two-country for a currency union, so some relationships are common?
6) Is it a problem of the model? (calculations etc)
Notice that the model from its nature is complicated but i cannot start from a simpler model, as it must have two countries and unemployment.

Please can somebody tell me what to do? ( how to solve this problem or what experienced people mostly do in this case)
If somebody needs to look into my paper, please ask me.

Thank you very much in advance.

Re: The rank condition ISN'T verified_1 colinear relationshi

PostPosted: Mon May 26, 2014 5:25 pm
by jpfeifer
Without knowing the model, it is hard to tell. Looking at the collinearity issue is tricky here, because your model has a unit root that will always result in a collinearity warning. My guess is that the indeterminacy is due to the model specification in the sense that the monetary policy rule in the home country together with the links to the other country are insufficient to uniquely pin down domestic and foreign inflation as well as the terms of trade changes. I would try whether the model solves for each country individually (autarky) and then focus on how to link the autark countries.

Re: The rank condition ISN'T verified_1 colinear relationshi

PostPosted: Mon May 26, 2014 7:22 pm
by nk178
Dear jpfeifer,

Thank you very much for your reply. I did what you told me and it works!!! I considered each country separately as autarky (so I ignored the terms of trades from the equations, also in the Fisher equation I substituted the cpi with the domestic price index, since the autarky economy does not import goods) and then I run dynare and it works for both countries! Therefore, it seems what you said is perfectly correct.

May I ask any suggestions from now on ?
Shall I leave the rest of the model as it is and try to find another specification for the interest rate rule? One maybe which affects the terms of trade? In general , for open economies can the terms of trade be included in the interest rule?

To be honest I chose that Taylor rule for my model in a kind of random way...I found some similar in the literature for open economies and I tried to use them.

If this is not the case may I ask for alternative suggestions?

Thank you very much for your reply. It is very useful.