# Correlations output for log-linearized model

**URL:** <https://forum.dynare.org/t/correlations-output-for-log-linearized-model/23296>\
**Category:** Stochastic simulations\
**Created:** [8 October 2023 18:32 UTC](https://forum.dynare.org/t/correlations-output-for-log-linearized-model/23296 "2023-10-08T18:32:45Z")\
**Posts on this page:** 3\
**Page:** 1

<div class="post-metadata">

**Author:** ![deniss](https://forum.dynare.org/letter_avatar_proxy/v4/letter/d/f9ae1b/32.png) [@deniss](https://forum.dynare.org/u/deniss)\
**Post date:** [8 October 2023 18:32 UTC](https://forum.dynare.org/t/correlations-output-for-log-linearized-model/23296/1 "2023-10-08T18:32:45Z")

</div>

Could anyone help please, I am a little confused.

If I write down my model in exp() in Dynare (i.e. log-linearized version), then the correlation table (and other moments) that Dynare prints are in levels? or gaps? For simulation procedure I use `stoch_simul` with no `hp_filter` option.

And correct me if I am wrong, but if I use one-sided HP filter for my data for Bayesian estimation purposes and compute moment, I should also simulate the model with one-sided HP filter option?  
In other words, to compare simulated and real moments, I need to have both real data and simulated data treated in the same way (one-sided filter)?

It would be extremely helpful to receive any suggestions. Thank you.

---

<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:** [10 October 2023 12:19 UTC](https://forum.dynare.org/t/correlations-output-for-log-linearized-model/23296/2 "2023-10-10T12:19:45Z")

</div>

1. If you did an exp()-substitution, everything will have an interpretation as being in percentages. Second moments in Dynare are centered moments, so the interpretation is always deviations from the mean, in this case percentage deviations from the mean.
2. Yes, meaningful comparisons usually involve applying the same filter to the model and the data.

---

<div class="post-metadata">

**Author:** ![deniss](https://forum.dynare.org/letter_avatar_proxy/v4/letter/d/f9ae1b/32.png) [@deniss](https://forum.dynare.org/u/deniss)\
**Post date:** [11 October 2023 13:30 UTC](https://forum.dynare.org/t/correlations-output-for-log-linearized-model/23296/3 "2023-10-11T13:30:26Z")

</div>

Many thank, Professor!