Page 1 of 1

STEADY: numerical initial values incompatible

PostPosted: Thu Sep 30, 2010 1:31 pm
by john
Hi, I am a beginner in Dynare and struggle with error messages.
It seems like my initial value is not relevant.
In this case, how can I get the steady state other than exactly provide them.

The error message is ....



STEADY: numerical initial values incompatible with the following equations
Columns 1 through 18

3 4 9 18 19 20 21 22 24 25 26 27 28 29 30 31 32 82

Columns 19 through 25

83 84 85 86 87 88 89

??? Error using ==> dynare_solve at 82
exiting ...

Error in ==> steady_ at 124
[oo_.steady_state,check] = dynare_solve([M_.fname '_static'],...

Error in ==> steady at 52
steady_;

Error in ==> ver01 at 586
steady;

Error in ==> dynare at 132
evalin('base',fname) ;

Re: STEADY: numerical initial values incompatible

PostPosted: Thu Sep 30, 2010 6:14 pm
by jsteinberg
It means that the initial values you provided do not satisfy some of the model equations (seems to be equations 1 - 25). You have to provide initial conditions that satisfy the system of equations that describes the model's equilibrium conditions.

Re: STEADY: numerical initial values incompatible

PostPosted: Thu Sep 30, 2010 8:15 pm
by john
Thanks a lot for your comment.

1) Does it mean I need to provide a exact steady state solution to these values?
2) Or simply can I provide them all zeros? (which satisfies the FOCs)?

If I need to provide a exact initial values.... it means I need to implement a steady state solver (?)
In this sense, Dynare provides only impluse response function other than steady state solution?

Sorry for the stupid quesion.....
Many thanks for your comment..

Re: STEADY: numerical initial values incompatible

PostPosted: Thu Sep 30, 2010 8:23 pm
by jsteinberg
You don't have to provide an exact solution, but you do need to provide an initial condition (consistent with the model equations, of course) from which the algorithm Dynares uses to find steady states will converge to the actual steady state. Starting with all zeros is unlikely to work. The algorithm uses derivatives, and you will likely have marginal utilities and marginal products that are infinite when real variables are zero. You should be able to guess at some of the steady state values by looking at the steady state versions of the model equations. For example, you can solve for the steady state value of capital in an RBC model using the Euler equation and the fact that labor supply should be about 0.3:
Code: Select all
1 = beta * (alpha*K^alpha*0.3^(1-alpha) + 1 - delta)

Obviously you can solve this by hand. From there you can provide guesses for the other values as functions of K and L (=0.3).