Dynare forums

Actually there is no linearized equation. gdp is defined as gdp=(1-k)*z*h, where z=1, k=(gamma/2)*pi^2. When computing the steady states, I have one equation for h_ss, one for R_ss and one for gdp_ss. It doesn't look like gdp_ss is redefined in the process, but maybe I have not understood what you really mean. Moreover, solving the nonlinear system with fsolve results in three complex roots for the steady states, even if I change the starting guess values.

Why do you have an equation for gdp_ss if it is a parameter?

Maybe I am confusing something. If you want to take a look, the reference paper is "Small and orthodox fiscal multipliers at the ZLB" by Braun, Koerber and Waki (2013). At page 6 and 7 you can find the Taylor rule, the process for gdp and the definition of resource cost of price adjustment, k; at page 8 the AS and AD schedules. These are the equations I wrote in the model block and from which I am trying to compute the steady states.

gdp_ss is definitely not a parameter, but a function of h_ss and pi_ss, that I fix equal to 0.02, the target rate implicitly assumed by the Taylor rule. So I computed the steady state values with an external Matlab program and I copied the resulting values in the initval-block. However the problem persists and I really don't know what's wrong, since the equations of the model are correct.

gdp_ss really is the steady state associated with the variable gdp. You can use the steady_state operator for this:

Code: Select all

gdp_hat=log(gdp)-log(steady_state(gdp));


Moreover, in New Keynesian models, the inflation rate is not endogenously determined but fixed to a particular value by the montary policy rule that sets the nominal interest rate. You need to set the monetary policy rule and the Euler equation in a consistent way to set target inflation to the desired level. Currently, you fix r_e to 1/beta, the real interest rate, implying that steady state inflation must be 0.

See the attached mod-file