st: esttab with xtreg (robust)

["Steinbach Sandro" <ssteinbach@ethz.ch>]

Hi,
I'm trying to create regression output tables with the esttab program. During a meeting with my supervisor, we found out that Stata reports different standard errors in the esttab output than if  the estimation is done in the Stata command line. To illustrate the problem, please run my  minimal working example:

--------------------------------------------------------------------------------------------------------------
*** simulating the performance of esttab with xtreg ***
*Stata  Version 13 MP2
*** set up dataset
clear all
set obs 1000
g i = _n
g y = rnormal()
expand 5
sort i
by i: g t=_n
g x = rnormal()
g u = rchi2(2)
by i: replace y = 0.8*y[_n-1] + 2*x + u if _n>1
xtset i t
***
* compare results for FE, robust
xtreg y x, fe robust
eststo: xtreg y x, fe robust
* and for the re model
xtreg y x, re robust
eststo: xtreg y x, re robust
* areg reports correct SE
areg y x,  absorb(i) robust
eststo: areg y x, absorb(i) robust
--------------------------------------------------------------------------------------------------------------

As you can see, the robust SE are larger when I save the model with eststo. I did some error tracking and found that the problem only occurs when FE or RE models are estimated with the robust command.

Does anyone has a clue why Stata reports different SE? (I guess it has something do with the version of Stata?)

I am looking forward to getting your answers.

Thank you,

Sandro Steinbach

--
ETH Zurich
Sandro Steinbach
Institute for Environmental Decisions
Agricultural Economics



*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/