# Inıtıal Conditions of the nonstationary variables cannot be estimated

**URL:** <https://forum.dynare.org/t/initial-conditions-of-the-nonstationary-variables-cannot-be-estimated/12495>\
**Category:** ML/Bayesian estimation\
**Created:** [27 September 2018 07:41 UTC](https://forum.dynare.org/t/initial-conditions-of-the-nonstationary-variables-cannot-be-estimated/12495 "2018-09-27T07:41:35Z")\
**Posts on this page:** 2\
**Page:** 1

<div class="post-metadata">

**Author:** ![ykkoca](https://forum.dynare.org/letter_avatar_proxy/v4/letter/y/aeb1de/32.png) [@ykkoca](https://forum.dynare.org/u/ykkoca)\
**Post date:** [27 September 2018 07:41 UTC](https://forum.dynare.org/t/initial-conditions-of-the-nonstationary-variables-cannot-be-estimated/12495/1 "2018-09-27T07:41:36Z")

</div>

Hi all,

While Dynare runs csminwel.m to find the mod of the distribution, before estimating, it returns the warning “There isn’t enough information to estimate the initial conditions of the nonstationary variables. The diffuse Kalman filter never left the diffuse stage”. The warning is given over some iterations, and is not given during some. I wonder what could be causing this warning to be returned. If necessary, I will upload the datafile as well.

Edit: Please note that the stochastic simulation of the model perfectly runs, but the problem is with diffuse Kalman filter.

Edit 2: I specified both the domestic and the foreign consumption in first difference form. Once I change these with the level form, both of them get stationary, and I no longer need to use diffuse\_filter, so that I get rid of this problem.

[sscale.mod](https://forum.dynare.org/uploads/default/original/2X/f/f65e51868345e99252f7bc4bac20af442a57b6be.mod) (7.7 KB)

Thanks in advance,  
yakup

---

<div class="post-metadata">

**Author:** ![jpfeifer](https://forum.dynare.org/user_avatar/forum.dynare.org/jpfeifer/32/5044_2.png) [@jpfeifer](https://forum.dynare.org/u/jpfeifer)\
**Post date:** [28 September 2018 08:47 UTC](https://forum.dynare.org/t/initial-conditions-of-the-nonstationary-variables-cannot-be-estimated/12495/2 "2018-09-28T08:47:59Z")

</div>

Usually that is not a reason to worry. It often happens that some parameter draws during mode-finding result in issues. It would be a problem if that warning message would occur for every parameter draw as it then would indicate a problem with the model instead of a pathological parameter draw