There are 3 eigenvalue(s) larger than 1 in modulus

[xusherry]

Hi,guys
i am a new comer in dynare, and i am doing my project. But i cannot run the dynare code. And here is my problem:

There are 3 eigenvalue(s) larger than 1 in modulus
for 5 forward-looking variable(s)

The rank condition ISN'T verified!

Can u help me figure it out? Thanks a lot!

[jpfeifer]

Are you sure you uploaded the correct file? When I run it, the parameters are not correctly defined, because you use

Code: Select all

yss

before it is defined
Also, your law of motion for capital looks strange.

Code: Select all

k=delta_i*(x(-1)+i(-1))+(1-delta_i)*k(-1);


Are you using time to build or why is i lagged by one period?

[xusherry]

Thank you very much, maybe i uploaded the wrong code ,now i have upload the right code. My program is similar to the Ian Christensen &Ali Dib(2008), but i separate the entrepreneurs into two sectors, one is state-owned sector, the other is unstate-owned sector.
I tried to change some parameters, but it still can't work out, still shows "There are 3 eigenvalue(s) larger than 1 in modulus
for 5 forward-looking variable(s)" . i'm very confused how to solve this problem and really hope you could help me with this.
Tank you again!!

[xusherry]

Are you sure you uploaded the correct file? When I run it, the parameters are not correctly defined, because you use

Code: Select all

yss

before it is defined
Also, your law of motion for capital looks strange.

Code: Select all

k=delta_i*(x(-1)+i(-1))+(1-delta_i)*k(-1);


Are you using time to build or why is i lagged by one period?

Thank you very much, maybe i uploaded the wrong code ,now i have upload the right code. My program is similar to the Ian Christensen &Ali Dib(2008), but i separate the entrepreneurs into two sectors, one is state-owned sector, the other is unstate-owned sector.
I tried to change some parameters, but it still can't work out, still shows "There are 3 eigenvalue(s) larger than 1 in modulus
for 5 forward-looking variable(s)" . i'm very confused how to solve this problem and really hope you could help me with this.
Tank you again!!

[jpfeifer]

Again my question: why are you using this strange timing for investment? The law of motion for capital, eq. (23) in Christensen/Dib is very different.

[xusherry]

Again my question: why are you using this strange timing for investment? The law of motion for capital, eq. (23) in Christensen/Dib is very different.

Dear professor jpfeifer,
The equation(23) is same as the equation in Christensen/Dib, maybe your question is about the eq(22), in eq(22), k means the capital, and it is the capital evolution equation, i put the time one period ago because in some examples i have learned ,they all put the capital or bond variable one time period ago.
Also i have tried to write the eq(22)the same as in Christensen/Dib, but the problem still exists.
Now i have changed eq(3) and eq(4) through writting the time one period ago,here is my newst code, and the warning information is coming up like this:
"There are 4 eigenvalue(s) larger than 1 in modulus
for 4 forward-looking variable(s)

The rank condition ISN'T verified!

错误使用 print_info (line 48)
Blanchard Kahn conditions are not satisfied: indeterminacy due to rank failure"

I don't know what is wrong on earth, is it a timing problem or a parameter calibration problem? I'm looking foward your helpful reply, thank you very very much !!!!!!

[jpfeifer]

Equation (23) of Christensen/Dib (2008), Review of Economic Dynamics, p. 155–178 is the law of motion for capital. The sentence preceeding it is

The aggregate capital stock evolves according to

You numbered it (22) in your mod-file, but it actually is equation (23) in the PDF of the paper. The equation in your mod-file reads

k=delta_i*(x(-1)+i(-1))+(1-delta_i)*k(-1);%(22) Capital evolution equation

This is simply wrong. You are correct in shifting the timing of capital. But the timing of the other variables, which are not predetermined, is not affected by this. Therefore, the correct Dynare implementation should be

Code: Select all

k=delta_i*(x+i)+(1-delta_i)*k(-1);


An additional complication is the two-sector structure, so first make sure the model works with one sector.