BK condition is verified, why the irfs do not go back zero?

[kang]

Hi, dear all,

I am running a dynare code to solve for a small open economy model, the bk condition is verified, but the irfs does not go back the zero. In fact, the irfs seem to be ok in the first 20 or 40 periods, but when i check more periods, they are not stable at all. I wonder what kind of problem it could be ? Anyone can help? The code is attached.

thanks

[jpfeifer]

Hi,

apparantly the BK-conditions are satisfied as you have 2 eigenvalues larger than 1 in modulus for 2 forward-looking variables. However, your third-largest eigenvalue is exactly 1, which should explain the permanent impulse responses as you have a unit root in your model. Maybe this is a problem of the unit root in consumption usually encountered in small open economy models. In this case you should see Schmitt-Grohe/Uribes "Closing small open economy models" (http://econ.duke.edu/~grohe/research/closing_jie.pdf) to close the model and render it stationary.

[kang]

Thanks for your reply and help. But in the model, I have added the bond adjustment cost, which is the psi_b*b in the second equation.

[jpfeifer]

I just recognized that your model does not feature depreciation of capital. This is the reason for the unit root. If you set your law of motion from

Code: Select all

k=k(-1)+i/kss;

to

Code: Select all

k=(1-delta)*k(-1)+i/kss;

with 0<delta<1, the impulse responses will be stationary.