# Parameter calibration
**URL:** https://forum.dynare.org/t/parameter-calibration/13469
**Category:** ML/Bayesian estimation
**Created:** [25 March 2019 10:14 UTC](https://forum.dynare.org/t/parameter-calibration/13469 "2019-03-25T10:14:03Z")
**Posts on this page:** 13
**Page:** 1
### Author:  [@hipa](https://forum.dynare.org/u/hipa)
#### Post date: [25 March 2019 10:14 UTC](https://forum.dynare.org/t/parameter-calibration/13469/1 "2019-03-25T10:14:03Z")
Dear members,
I am encountering an issue in my calibration exercise. After rearranging the CPOs, I obtain for my parameter the following equation,
My issue is with “sigma es”. All ratios and parameters present in the equation are already known, except for “sigma es”, which is a parameter I would like to estimate via bayesian estimation. That “sigma” represents an elasticity of substitution.
Can I specify to the “parameter” block in dynare the equation for the parameter “alpha d”, so that when it computes “sigma es” via the estimation block, dynare computes from the equation the final value for “alpha d” from the value of “sigma es”?
Or do I have to impose a value to “sigma es” in order to compute the value of “alpha d”?
Thank you in advance.
Best,
---
### Author:  [@jpfeifer](https://forum.dynare.org/u/jpfeifer)
#### Post date: [27 March 2019 19:37 UTC](https://forum.dynare.org/t/parameter-calibration/13469/2 "2019-03-27T19:37:03Z")
What you describe sounds like a case of parameter dependence. You should define \alpha\_d as a model-local variable with the `#`-operator and then estimate \sigma\_{es}
---
### Author:  [@hipa](https://forum.dynare.org/u/hipa)
#### Post date: [28 March 2019 08:41 UTC](https://forum.dynare.org/t/parameter-calibration/13469/3 "2019-03-28T08:41:51Z")
Thank you very much.
Is normalization compulsory when having CES functions? I’ve seen several authors saying that it is when one wants to compare two economies.
Thanks,
---
### Author:  [@jpfeifer](https://forum.dynare.org/u/jpfeifer)
#### Post date: [30 March 2019 14:13 UTC](https://forum.dynare.org/t/parameter-calibration/13469/4 "2019-03-30T14:13:59Z")
Have a look at the Cantore/Levin (2012) paper on [Getting normalization right: Dealing with ‘dimensional constants’ in macroeconomics](http://dx.doi.org/10.1016/j.jedc.2012.05.009)
---
### Author:  [@hipa](https://forum.dynare.org/u/hipa)
#### Post date: [1 April 2019 11:50 UTC](https://forum.dynare.org/t/parameter-calibration/13469/5 "2019-04-01T11:50:49Z")
Would you recommend to use net national product rather than gross national product to compute steady state ratios and factor’s share ?
Thank you.
Best,
---
### Author:  [@jpfeifer](https://forum.dynare.org/u/jpfeifer)
#### Post date: [4 April 2019 09:40 UTC](https://forum.dynare.org/t/parameter-calibration/13469/6 "2019-04-04T09:40:16Z")
That is impossible to answer in general. Model and data should be consistent. Thus, you need to use the suitable data concept.
---
### Author:  [@hipa](https://forum.dynare.org/u/hipa)
#### Post date: [11 April 2019 08:43 UTC](https://forum.dynare.org/t/parameter-calibration/13469/7 "2019-04-11T08:43:42Z")
Good morning,
Is it ok if alpha\_d depends not only on other parameters (sigma\_es, beta, deal\_d) as well as endogenous variables (e\_h and d) and exogenous variables (p\_e and t\_e t\_c)? From what I saw, most codes uses the #-operator with an equation linking parameters to other parameters (and not exo and endogenous variables).
THANKS
---
### Author:  [@jpfeifer](https://forum.dynare.org/u/jpfeifer)
#### Post date: [11 April 2019 08:57 UTC](https://forum.dynare.org/t/parameter-calibration/13469/8 "2019-04-11T08:57:08Z")
If it depends on a variable, it is not a parameter.
---
### Author:  [@hipa](https://forum.dynare.org/u/hipa)
#### Post date: [11 April 2019 08:59 UTC](https://forum.dynare.org/t/parameter-calibration/13469/9 "2019-04-11T08:59:17Z")
Right. But I have nested CES functions that gives me at SS parameters depending upon variables … I’m going to try to simplify more the equations then.
Thanks
---
### Author:  [@jpfeifer](https://forum.dynare.org/u/jpfeifer)
#### Post date: [11 April 2019 09:03 UTC](https://forum.dynare.org/t/parameter-calibration/13469/10 "2019-04-11T09:03:02Z")
A variable in steady state is constant and only depends on parameters. Therefore it is not a variable anymore. If your parameter depends on a steady state value, that is fine.
---
### Author:  [@hipa](https://forum.dynare.org/u/hipa)
#### Post date: [11 April 2019 09:07 UTC](https://forum.dynare.org/t/parameter-calibration/13469/11 "2019-04-11T09:07:05Z")
Yes, when I was saying “endogenous variables” or “exogenous” I meant at steady state. The above equation is at SS.
Should I specify SS variables as well?
#alpha\_d= 1/(beta\*(1+t\_c)_p\_e_(1+t\_e)_(e\_h\_ss/d\_ss)^(1/sigma\_es)+beta_(1-delta\_d)+1
// In my code I have declared in the var-block: e\_h and d. Should I add as well e\_h\_ss and d\_ss or can I just put the following equation:
#alpha\_d= 1/(beta\*(1+t\_c)_p\_e_(1+t\_e)_(e\_h/d)^(1/sigma\_es)+beta_(1-delta\_d)+1
But I’m afraid in that case Dynare will take e\_h and d as non SS values
Thanks
---
### Author:  [@jpfeifer](https://forum.dynare.org/u/jpfeifer)
#### Post date: [11 April 2019 10:08 UTC](https://forum.dynare.org/t/parameter-calibration/13469/12 "2019-04-11T10:08:43Z")
Either define parameters that store the steady state values or use the `STEADY_STATE()`-operator.
---
### Author:  [@hipa](https://forum.dynare.org/u/hipa)
#### Post date: [11 April 2019 11:15 UTC](https://forum.dynare.org/t/parameter-calibration/13469/13 "2019-04-11T11:15:57Z")
Thank you.
Have a nice day