st: F-test with robust standard errors

[Veronika <angela.djupsjobacka@abo.fi>]

Dear statalisters,

I would like to test the hypothesis that a group of parameters jointly equal
zero. 

This is the setting of the problem: I have a panel of firms that I follow
over time, measuring yearly output. I also identify managers changing
between firms. This is a simulated data that I use to try out the methods
before I start working on the real data. I have constructed the data so that
it consists of firm effects, year effects and heteroskedasticity between
firms, but no effects of managers. I want to test H0 that the effects of
managers jointly equal zero (and H0 is true). I use the following command
(the third line giving the test results of special interest to me):

xtset firm year
xtreg output year2-year20 manager2-manager100, fe vce(robust)
testparm manager2-manager100

However, the results show that managers have a highly significant
"explaining power". I have resampled many times and I can always reject H0.
This is not the case if i don't use the "vce(robust)"-option (then
F-statistic and p-values are ok). 

So my question is: What command should I use to test the hypothesis that the
manager effects equal zero, if I still want to have robust standard errors
(and F-statistic)? Appearently, the testparm-command doesn´t give me what
I'm searching for...

Thanks!

Veronika

 
   

--
View this message in context: http://statalist.1588530.n2.nabble.com/F-test-with-robust-standard-errors-tp6193538p6193538.html
Sent from the Statalist mailing list archive at Nabble.com.

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