Esto with autoregressive residuals by groups

[Harshal Chaudhari]

Hi,

I am using xtmixed with residual structure as residuals(ar 1, by(rank_category) t(week)) with variance flag set. The xtmixed output gives a separate var(e) for each grouping level ie. rank_category. While using estout, I am able to calculate the var(e) for each level (I figured that stata paramaterizes it relative to first level of grouping.) However, I am still unable to reproduce the standard error associated with this variance in my esttab. The command I am using is as follows :

esttab, se wide transform(ln*: exp(2*@) 2*exp(2*@) at*: tanh(@) (1-tanh(@)^2) r_atr*: tanh(@) (1-tanh(@)^2) r_ln*: exp(2*(@+[lnsig_e]_b[_cons])) 2*exp(2*(@+[lnsig_e]_b[_cons])))

The output stub from xtmixed is as follows :

------------------------------------------------------------------------------
Random-effects Parameters | Estimate Std. Err. [95% Conf. Interval]
-----------------------------+------------------------------------------------
-----------------------------+------------------------------------------------
Residual: AR(1), |
by rank_category |
1: rho | .4616144 .0808769 .2894086 .6048761
var(e) | .18252 .0285839 .1342784 .248093
|
2: rho | .7508995 .0442481 .6505045 .8255044
var(e) | .5935972 .0977483 .4298565 .8197099
|
3: rho | .3634431 .0763827 .2053918 .5030186
var(e) | .1841135 .0194416 .1496933 .2264482
------------------------------------------------------------------------------

While my esttab command gives output as :

-----------------------------------------
lnsig_e
_cons 0.183*** (0.0286)
-----------------------------------------
r_atr1
_cons 0.462*** (0.0809)
-----------------------------------------
r_lns2ose
_cons 0.594*** (0.134)
-----------------------------------------
r_atr2
_cons 0.751*** (0.0442)
-----------------------------------------
r_lns3ose
_cons 0.184 (0.0368)
-----------------------------------------
r_atr3
_cons 0.363*** (0.0764)
-----------------------------------------


Any help would be appreciated, thanks!