Dear Statalisters
I am new to using IV regression.
I estimated a random effects GLS equation
-xtreg Y x1 x2 x3 x4 x5-10, re
I believe x1=own training is endogenous
x2, x3 and x4 are endogenous. The rest I believe are
not. I am using cross-section data and x1 is binary.
Do I have this variable as x1a if they have had
training and x1b (where I set x1b=0 if they have had
no training) if they have not had training.
Is it best to deal with one endogenous variable at a
time.
(2) I am not certain as to the best way to deal with
it either.
Does one estimate a probit thus:
-probit x1 x2 (without putting the instrument for
training) x3 x4
then
-predict x2hat
-xtivreg Y (x2=instrument for training, bwd) x2hat x3
x4 ..., re
The instrument for training, bwd was created as
follows:
bwd=.
bwd=1 if they have had training
bwd=0 if they have not had training
(3) Is it better to estimate xtivreg in my case,
rather than -ivreg-.
(4) Should I then look if bwd is significant. Does the
sign matter. A priori would one expect the sign to be
negative if this variable is not endogenous to the
dependent variable.
(5) Would expect the own training variable to be
negative and significant if training is not endogenous
to Y.
(6)Then does one do the
-hausmen wu- test. If I want training to be
non-endogenous to Y, what would be my a priori
expectations.
I would be grateful, if you would confirm the above.
Thanks in advance
Renuka
___________________________________________________________
Yahoo! Answers - Got a question? Someone out there knows the answer. Try it
now.
http://uk.answers.yahoo.com/
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
