Stochastic steady-state in Dynare++

[ivan_lozev]

I am currently trying to use Dynare++ to replicate the results on Burnside (1998) model, discussed in:
Juillard, Michel. "Local approximation of DSGE models around the risky steady state." Wp. comunite 87 (2011).
To obtain the steady state I use the command

dynare++ --steps 5 burnside.mod, and then retrieve the sequence of steady-states (in Matlab) by the command
dyn_steady_states.
Unfortunately, the entire sequence just gives the deterministic steady state.

Thank you,
Ivan Lozev
PhD student Hamburg University

Below I enclose the burnside.mod model:

var y, x;
varexo e;

parameters beta, rho, x_bar, theta, sigm;
beta = 0.95;
rho = -0.139;
x_bar = 0.0179;
theta = -10;
sigm = 0.0348;

model;
y = beta*exp(theta*x(+1))*(1+y(+1));
x = (1-rho)*x_bar+rho*x(-1)+sigm*e;
end;

initval;
x = x_bar;
y = 5;//beta*exp(theta*x_bar)/(1-beta*exp(theta*x_bar));
end;

vcov = [1];

order = 1;

[ivan_lozev]

It has been 10 days already without any reply. Probably not many people use Dynare++ to calculate stochastic steady-states. Or am I unclear in my post above?

[jpfeifer]

Sorry, Michel did not yet find the time to reply. If I read p. 9 of the Kamenik (2011): "DSGE Models with Dynare++. A Tutorial." correctly, then

Code: Select all

dyn_steady_states

provides the deterministic steady state, while

Code: Select all

dyn_ss

should provide the fixed point of the decision rules, commonly called the stochastic steady state.

[ivan_lozev]

Thank you for your reply, Johannes. It seems that the --steps command computes values in <dyn_steady_states> different from the deterministic steady-state only when in the mod file the order of approximation is more than one. Then dyn_ss also gives values closer to the exact solution, rather than just the deterministic steady-state. I'll probably use the dyn_ss with order=2 in the mod file and not run any --steps command.
Thanks again,
Ivan

[jpfeifer]

I don't get your first point. If the approximation order is 1, then the model is certainty equivalent and the stochastic steady state is equal to the deterministic steady state.

[ivan_lozev]

That's right - first order approximation is certainty equivalent. But you can still, in principle, do first-order approximation around the stochastic steady-state, rather than the deterministic one. That is what de Groot, Oliver. "Computing the risky steady state of DSGE models." Economics Letters 120.3 (2013): 566-569. does, for example.

[jpfeifer]

That is a very different approach. In that paper, you use a second order approximation to get the stochastic steady state and then do a first order approximation around this point. Therefore, the approximation orders involved are "inconsistent". Dynare++ uses order=1 consistently, implying that the deterministic and stochastic steady state will coincide at order=1.

[salga010]

Hi all

Maybe somebody provided a solution for the following problem. I am trying to run Dynare++ in MAC OX El Capitan but it is giving the following error

====
x-134-84-1-73:Dynare$ /Applications/Dynare/4.4.3/dynare++/dynare++ example1.mod
dyld: lazy symbol binding failed: Symbol not found: _dgetrf_
Referenced from: /Applications/Dynare/4.4.3/dynare++/dynare++
Expected in: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib

dyld: Symbol not found: _dgetrf_
Referenced from: /Applications/Dynare/4.4.3/dynare++/dynare++
Expected in: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib

Trace/BPT trap: 5
===

Seems that the issue has to do with libBLAS to be incompatible in El Capitan, or even Sierra OS. Is there any way ti sile this problem? There are some tutorial online but involves copy an old version of libBLAS which I would prefer not to do.
Thanks!!