Statalist


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

st: fe questions


From   sara borelli <saraborelli77@yahoo.it>
To   statalist@hsphsun2.harvard.edu
Subject   st: fe questions
Date   Fri, 23 Oct 2009 10:09:26 +0000 (GMT)

Dear All,

I have a couple of questions. I have found something similar in older posts but I have not clear the answer. Thank you in advance for any help

I want to estimate the following model by fixed effects
Yjt = b*Xjt + Ujt
Yjt = birth rate in county j in year t
Xjt county level covariates

there are 615 counties and 8 years

Question 1
I estimated the model by using
1) xtreg Yjt Xjt, fe i(county)cluster(county)
2) xi:  reg Yjt Xjt i.county, cluster(county)
3) areg Yjt Xjt, absorb(county) cluster(county)
The coefficients estimates are the same in the 3 specifications.
The standard errors are exactly the same in 2) and 3) (as it should be)
but slightly different fron 1). I understand that there is some difference in the degrees of freedom adjustment, but I am not sure...

Question 2
I re-estimated the model by weighting each observation by the county population in each year, POPjt
1) xtreg Yjt Xjt [aweight=POPjt], fe i(county)cluster(county)
2) xi:  reg Yjt Xjt i.county [aweight=POPjt], cluster(county)
3) areg Yjt Xjt [aweight=POPjt],  absorb(county) cluster(county)
agai, 2) and 3) give exactly the same result, but 1) cannot be estimated because STATA says that weights must be constant whithin panel.
I do not understand this and why xtreg cannot estimate the weighted regression while the dummy variable regression allows weights

thank you for any help
Sara
 

 







      

*
*   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/



© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index