Second-order cross partial error


#1

I am trying to simulate a 2-good version of a model that I’ve successfully simulated and estimated in a 1-good version. “rho” is an elasticity parameter for a Dixit-Stigitz type of model that has to exceed one. For some reason, in this 2-good version (attached), if I set rho to anything else but 2 (even 2.1) I get the error message

“the second-order cross partial of equation 3 w.r.t. trend variable k and endogenous variable h1 is not null.”

In my experience, that is a message that usually means either the model is not homogeneous of degree one in the trending variable, or some other variable is trending that wasn’t listed among the deflated variables. I’ve checked and double-checked, and everything seems right, but I keep getting this error message. Specifically, equation 3 is just a CES combination of two trending variables with elasticity rho. I’ve attached the .mod file, would appreciate any help. Thanks.
model2work1_highgam.mod (3.42 KB)


#2

Hi Forum,

I also have a rather quick question regarding this warning messages. What are the effects of such a warning on the results? I ran a small toy model just to learn about the trend_var command. First without errors, but when I inserted the calculated steady state into initval; dynare produced such warnings:

“trends not compatible with balanced growth path; the second-order cross partial of equation 7 (line 52) w.r.t. trend variable A and endogenous variable K is not null”

What to do? I am rather clueless… Any help is much apprechiated!
I include the mod file.
test_growth.mod (567 Bytes)


#3

Your code pretty much looks like the example at https://git.dynare.org/Dynare/dynare/blob/master/examples/fs2000_nonstationary.mod
So my hunch is that the error is due to a persistent problem with that part of Dynare https://github.com/DynareTeam/dynare/issues/1389


#4

Thank you for the hint!