# Bayesian estimation yet

**URL:** <https://forum.dynare.org/t/bayesian-estimation-yet/12040>\
**Category:** ML/Bayesian estimation\
**Created:** [25 June 2018 02:34 UTC](https://forum.dynare.org/t/bayesian-estimation-yet/12040 "2018-06-25T02:34:19Z")\
**Posts on this page:** 7\
**Page:** 1

<div class="post-metadata">

**Author:** ![fabianowar](https://forum.dynare.org/letter_avatar_proxy/v4/letter/f/f05b48/32.png) [@fabianowar](https://forum.dynare.org/u/fabianowar)\
**Post date:** [25 June 2018 02:34 UTC](https://forum.dynare.org/t/bayesian-estimation-yet/12040/1 "2018-06-25T02:34:19Z")

</div>

Dears users, I have a problem with Bayesian estimation of my model, since I can only estimate with the inclusion of only one Bstar variable, what is happening that I can not use the others?

Best regards

Thanks! [openeconomy.zip](https://forum.dynare.org/uploads/default/original/2X/5/5f6bb20bf8f61c56c4cfd2a3a25f1b994db5b39f.zip) (21.1 KB)

---

<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:** [25 June 2018 18:10 UTC](https://forum.dynare.org/t/bayesian-estimation-yet/12040/2 "2018-06-25T18:10:20Z")

</div>

It’s hard to tell. The fact that adding any other observable leads to issues indicates a numerical problem rather than standard stochastic singularity. Are you sure your data treatment is correct? Your data features an extremely large standard deviation. `Y` is between -40 and +40. With the usual shock sizes, this is hard to capture.

---

<div class="post-metadata">

**Author:** ![fabianowar](https://forum.dynare.org/letter_avatar_proxy/v4/letter/f/f05b48/32.png) [@fabianowar](https://forum.dynare.org/u/fabianowar)\
**Post date:** [25 June 2018 18:14 UTC](https://forum.dynare.org/t/bayesian-estimation-yet/12040/3 "2018-06-25T18:14:56Z")

</div>

Yes, porfessor! The HP filter was used. So I do not know what else I can do to make the variance of the data less variant.

---

<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:** [25 June 2018 18:17 UTC](https://forum.dynare.org/t/bayesian-estimation-yet/12040/4 "2018-06-25T18:17:15Z")

</div>

1. You should not use the two-sided HP filter for estimation. See [Demean the series or not](https://forum.dynare.org/t/demean-the-series-or-not/11130)
2. Did you take the log before filtering for growing variables? And was the underlying data seasonally adjusted?

---

<div class="post-metadata">

**Author:** ![fabianowar](https://forum.dynare.org/letter_avatar_proxy/v4/letter/f/f05b48/32.png) [@fabianowar](https://forum.dynare.org/u/fabianowar)\
**Post date:** [25 June 2018 18:20 UTC](https://forum.dynare.org/t/bayesian-estimation-yet/12040/5 "2018-06-25T18:20:37Z")

</div>

I did not get the log before, but they were adjusted seasonally. So, would you advise me to take the log and then adjust seasonally?

---

<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:** [25 June 2018 18:23 UTC](https://forum.dynare.org/t/bayesian-estimation-yet/12040/6 "2018-06-25T18:23:25Z")

</div>

That explains it. You need to make sure the scaling of your model variables and the data are consistent. That is currently not the case. Please read my Guide to Observation Equations.

---

<div class="post-metadata">

**Author:** ![fabianowar](https://forum.dynare.org/letter_avatar_proxy/v4/letter/f/f05b48/32.png) [@fabianowar](https://forum.dynare.org/u/fabianowar)\
**Post date:** [25 June 2018 18:24 UTC](https://forum.dynare.org/t/bayesian-estimation-yet/12040/7 "2018-06-25T18:24:55Z")

</div>

> [@jpfeifer](#):
>
> That explains it. You need to make sure the scaling of your model variables and the data are consistent. That is currently not the case. Please read my Guide to Observation Equations.

Ok, professor! Really thanks.
