Efficient quantity


By now you should know the drill: without the mod-file we cannot help you!


Dear professor
I sent on your Email, because it must be secret.


model_diagnostics says

MODEL_DIAGNOSTICS:  The Jacobian of the static model is singular
MODEL_DIAGNOSTICS:  there is 1 colinear relationships between the variables and the equations
Colinear variables:
Colinear equations
  Columns 1 through 19

     1     2     3     5     6     7     8     9    10    11    12    13    14    15    16    17    18    19    20

  Column 20


MODEL_DIAGNOSTICS:  The presence of a singularity problem typically indicates that there is one
MODEL_DIAGNOSTICS:  redundant equation entered in the model block, while another non-redundant equation
MODEL_DIAGNOSTICS:  is missing. The problem often derives from Walras Law.


tanks a lot
I should delete g or t variable from my model?


Your are the model builder. You need to understand why g and t are collinear.


tanks a lot professor


Dear professor
after doing your suggestion in my model, model_diagnostics says
but applying identification says:
"There are NaN’s in the theoretical moments: make sure that for non-stationary models stationary transformations of non-stationary observables are used for checking identification. [TIP: use first differences]."
what is the problem for my error in identification? and does model_diagnostics says the model is suitable?


This indicates that one of your observables defined with varobs has a unit root, which is quite uncommon.


Dear professor
my model is Log-Linearized and all of the model variable has defined by deviation from steady-state and initial value of all variable is zero , but observable variable has defined by log-difference.
my question is
1-how coding the measurement equation in dynare? supposing “y” is model variable and “yo” is observed variable.


This has nothing to do with the problem. An observed log-difference should be stationary. The error message suggests that at least one of the observed variables is non-stationary. Regarding the coding of measurement equation, I already referred you to my Guide in the above posts.


Dear professor
I want to estimate potential output by Kalman smoother, when I use “smoother” in “estimation” command.
the result of Kalman smoother and observed variable is identical.
what is the problem?
tanks a lot for your time


There is nothing wrong. Observables are observed, so there is no uncertainty about their values. Unless you have stochastic singularity or measurement error, the smoothed observable will be identical to the actual values.


tanks a lot dear professor
then how to create a difference between it for indication output gap.
in fact how coding the potential output?


As @stepan-a said

You need to have a model that define the concept of an output gap (which is by definition unobservable). If you model does not have that, there is nothing you can do with it.


Dear professor
I appreciated for your time
I have one problem and yet dose not solved.
i have y variable in the model and i have yo as observable variable, yo defined as “yo=d(log(Y))”.
my question: do i define measurement equation as “yo=y-y(-1)”? or “yo=y-y(-1)-epz” ?
epz is growth rate of technology.
tanks a lot


As discussed in my Guide, if technology has a unit root, the growth rate definition in the model will feature the growth rate of technology. If technology is stationary, then y will already capture technology shocks.


tanks a lot
technology growth rate does not has unit root. then which one equation do I use?
“yo=y-y(-1)”? or “yo=y-y(-1)-epz”?


Obviously the first one.


Dear professor
how can i graph an irf for multi shock and an endogenous variable in one graph?
tanks a lot


After running the mod file, you can get the irf results in oo_.irfs. Then you can plot them by matlab yourself. Actually, in the folder named ‘doc’ under root directory of dynare you can find the reference manual and user guide, make good use of them. Hope this helped.