Transforming random-effects parameters of an xtmixed model with two levels

[Agustina Malvido]

Hi,
I have a two Level xtmixed model I can't transform properly the random-effects Parameters to get them as standard deviations in esttab.

This is my model:
xtmixed logrealpricehl realmonthly_av_price_must_hl prov_wine_elaborated_10000hl percentage_change_must_2011_02 av_hl_prodcoopiofs2 red ownbrand coop_buyer swshare_of_producers, || region: ||cod_dept_buyer: , mle

I wrote:
esttab reg1 using regwine_shares156633.rtf, star(* 0.10 ** 0.05 *** 0.01) r2 ar2 onecell nogaps scalars("F" "chi2_c") wide transform(ln*: exp(@) exp(@)) eqlabels("" "sd(_cons)" "sd(_cons)" "sd(Residual)", none) varlabels(,elist(region:_cons "{break}{hline @width}")) varwidth(13) replace b(%10.4f) se
But I don’t get all the standard deviations. I think I have a problem addressing the two levels.
Anyone has some advice about it?
Best,
Agustina

[Scott Merryman]

Here is an example obtaining the standard deviation of the the random effects in esttab (written by Ben Jann, SJ14-2 Making regression tables from stored estimates, findit esttab to locate and download).

. webuse productivity
(Public Capital Productivity)

. eststo: mixed gsp private emp hwy water other unemp || region: || state: , stddev noheader nogroup nofet nolog

------------------------------------------------------------------------------
  Random-effects Parameters  |   Estimate   Std. Err.     [95% Conf. Interval]
-----------------------------+------------------------------------------------
region: Identity             |
                   sd(_cons) |    .038087   .0170591      .0158316     .091628
-----------------------------+------------------------------------------------
state: Identity              |
                   sd(_cons) |   .0792193   .0093861      .0628027    .0999273
-----------------------------+------------------------------------------------
                sd(Residual) |   .0366893    .000939      .0348944    .0385766
------------------------------------------------------------------------------
LR test vs. linear model: chi2(2) = 1154.73               Prob > chi2 = 0.0000

Note: LR test is conservative and provided only for reference.
(est1 stored)

.  esttab, onecell nogaps not  transform(ln*: exp(@) exp(@))

----------------------------
                      (1)   
                      gsp   
----------------------------
gsp                         
private             0.267***
emp                 0.754***
hwy                0.0710** 
water              0.0761***
other             -0.1000***
unemp            -0.00590***
_cons               2.129***
----------------------------
lns1_1_1                    
_cons              0.0381***
----------------------------
lns2_1_1                    
_cons              0.0792***
----------------------------
lnsig_e                     
_cons              0.0367***
----------------------------
N                     816   
----------------------------
* p<0.05, ** p<0.01, *** p<0.001

[Gozde Corekcioglu]

Hi,
I am having a problem transforming the unstructured covariance from xtmixed. The code works when I transform the rest of the variance components, but not when I also transform the covariance.

esttab model1 model2, mtitles("Model 1" "Model 2") ///
se transform(ln at : exp(@)^2 exp(@)^2 exp(@)^2 exp(@)*exp([lns1_1_1]_b[_cons])*exp([lns1_1_2]_b[_cons]) )

Thanks for your help!

[Oded Mcdossi]

You can find here some examples for esttab with mixed command. I will replicate Scott's example.

.  est clear

.  eststo: mixed gsp private emp hwy water other unemp || region: || state: hwy,cov(uns) stddev noheader nogroup nofet nolog   

------------------------------------------------------------------------------
  Random-effects Parameters  |   Estimate   Std. Err.     [95% Conf. Interval]
-----------------------------+------------------------------------------------
region: Identity             |
                   sd(_cons) |    .027124   .0831532      .0000667    11.03705
-----------------------------+------------------------------------------------
state: Unstructured          |
                     sd(hwy) |   .2771839   .0407205      .2078356    .3696716
                   sd(_cons) |   2.558362   .3679016      1.929995    3.391312
             corr(hwy,_cons) |  -.9984721   .0007565     -.9994212   -.9959698
-----------------------------+------------------------------------------------
                sd(Residual) |   .0302653   .0008933      .0285641    .0320678
------------------------------------------------------------------------------
LR test vs. linear regression:       chi2(4) =  1294.89   Prob > chi2 = 0.0000

Note: LR test is conservative and provided only for reference.
(est1 stored)

.  esttab, onecell nogaps not transform(ln*: exp(@) exp(@) at*: tanh(@) (1-tanh(@)^2))

----------------------------
                      (1)   
                      gsp   
----------------------------
gsp                         
private             0.180***
emp                 0.858***
hwy                0.0514   
water              0.0715***
other             -0.0810***
unemp            -0.00449***
_cons               2.359***
----------------------------
lns1_1_1                    
_cons              0.0271   
----------------------------
lns2_1_1                    
_cons               0.277***
----------------------------
lns2_1_2                    
_cons               2.558***
----------------------------
atr2_1_1_2                  
_cons              -0.998***
----------------------------
lnsig_e                     
_cons              0.0303***
----------------------------
N                     816   
----------------------------
* p<0.05, ** p<0.01, *** p<0.001