I have one order indeterminacy issue,id est :
“There are 4 eigenvalue(s) larger than 1 in modulus
for 5 forward-looking variable(s).” is the warning message I get
This model works well in Uhlig toolkit, but unfortunately it does not work in dynare. I lagged UIP condition or Euler eq. then it worked, but my professor says that this “trick” is wrong, as you are defining expectational equation as predetermined one.
I am attaching codes and data file
Any comments are welcome, as I am approaching to thesis deadline…
Try running your mod file, then placing a break-point at line 367 of dr1.m, running e.g. “check;” and looking at the value of condest(w1) when the break-point is hit. It may be marginal.
Having run your model the problem is more severe than I thought it was. I expect the equation
is wrong. Are you sure you’ve got your timing protocol right. t-dated variables are known at t. So if something is a pre-determined variable in the B-K sense it should have a (-1).
[quote=“cfp”]Having run your model the problem is more severe than I thought it was. I expect the equation
q(+1)=q-r+pi_t(+1)+r_st+nu_q;
is wrong. Are you sure you’ve got your timing protocol right. t-dated variables are known at t. So if something is a pre-determined variable in the B-K sense it should have a (-1).
Tom[/quote]
Dear Tom
Thanks for reply. I think the equation is right, I checked the paper again, and it seems timing format is ok. Further more I am ataching Lubik et al paper’s code they have the same format in UIP condition:
// 12. UIP equation
R - R_star = PI(+1) - PI_star(+1) + S(+1) -S;
Note that S=-q in my model. Actually Lubik’s model is 2 country model, mine is small open economy,so I reduced it according to other paper. Please have a look on Lubik’s code. Maybe it will help.
Thanks in advance
Not sure where the problem is sorry. My suggestion would be to go back to a model that you know works, and change it one step at a time towards the model you’re interested in.
