st: xtivreg,re - test for heteroskedasticity, serial correlation and multicollinearity

["Daniel Schalling" <daniel.schalling@uni-weimar.de>]

Dear Statalist,

I did a xtivreg,re with 75 groups, 60 time intervals (days), 90 exogenous
variables and 1 endogenous variable with 1 instrument. After estimating the
model I want to check the assumptions. In fact I want to do a test for
a) heteroskedasticity
b) serial correlation
c) multicollinearity

I was looking for some solutions doing this and found a lot of tips, but it
seems like nothing really works for me.

a) heteroskedasticity:
- robvar - to test for groupwise heteroskedasticity (Stata tip 38) --> but
this only work with regress, not with instruments
- xtgls - but this only seems to work with xtreg, without instruments -->
http://www.stata.com/support/faqs/stat/panel.html

b) serial correlation
- abar - but this works just for an instrumental regression (ivreg2) -->
there are no panel information included
- xttest1 - just works after xtreg --> no instruments are included
- xtserial - just works after xtreg --> no instruments are included

c) multicollinearity
- vif - but this just works after regress --> it doesn't care about
endogenous variables and instruments

I really would appreciate any advice or tip for this matters.
Thank you very much, Daniel.

*
*   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/