Question about replication of Cao and Nie (2017)

Perfect foresight simulations
1

Hi all,

I am trying to replicate Cao and Nie (2017) with dynare.

I have pinned down the steady state. But when I add the positive TFP shock, A, into the model, the perfect sight solver cannot solve for it. The code (attached: “cn17_ver3.mod”) describing the shock is: "
initval;
A = 1;
end;
steady;

shocks;
var A;
periods 1 2:20;
values 1.0 1.03;
end;"

The error message says that, “SOLVE: maxit has been reached
Simulation of the perfect foresight model failed!Switching to a homotopy method…
Homotopy not implemented for purely forward models!
Failed to solve the model!
Return with empty oo_.endo_simul.”.

One thing I want to mention about Cao and Nie (2017) is that the production is using the contemporaneous land (h_t) other than landholding from last period (h_{t-1}), which is the model they simplified from Iacoviello (2005). To rule out the default setting of timing convention about state variable in dynare, I converted the land in production from h_t to h_{t-1}, and consequentially, the change of optimality conditions and steady states. Yet I cannot find a steady state. Attached “cn17_ver6.mod” is the code.

Would you please shed some light with me if you see where I did wrong?

Thank you in advance for you time and help.

cn17_ver3.mod (2.2 KB)
cn17_ver6.mod (3.0 KB)

2
  1. You have a singularity problem.
  b + m*h*q(+1) = 0; //assume binding

makes

  0 = mu* (b + m*h*q(+1));

redundant.
2. The timing does not affect the steady states. You can verify that your analytical steady state results in a residual in equation 9. You need to find out why.

3

Hi Professor Pfeifer,

Thank you so much for your help! I change the equation with the budget constraint and it is running. I decide to stick with the original timing. Thank you for the tip for the second question.

Have a good day.