Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: st: RE: regress command with panel data
From
Austin Nichols <[email protected]>
To
[email protected]
Subject
Re: st: RE: regress command with panel data
Date
Tue, 22 Mar 2011 18:06:29 -0400
Tunga Kantarcı <[email protected]> :
In other words, yes, the plain "regress y x" command gives you pooled OLS.
But you might want to use cluster-robust standard errors (cluster by
panel) in both your FE and OLS models.
On Tue, Mar 22, 2011 at 4:56 PM, Nick Cox <[email protected]> wrote:
> I don't know if this answers your question, but typing
>
> . regress y x
>
> doesn't give Stata any information on the structure of the data. Nor does -regress-, in this case, pay attention to other information that you have given Stata.
>
> The data are just a bundle of y's and x's so far as -regress- is concerned. You can think of it as panels or cross-sections, but it doesn't care either way.
>
> Otherwise put, you can repeat -regress- after
>
> . tsset, clear
>
> or
>
> . xtset, clear
>
> to check that it makes no difference what you say about panels.
>
> The plain -regress- will not be the "correct" analysis if there is any kind of dependence structure other than a linear relation between y and x. I'm puzzled by any implication otherwise. I guess that plenty of other people were puzzled too, which is why you got no replies first time round.
>
> Nick
> [email protected]
>
> Tunga Kantarci
>
> I asked it before but could not get any reply. I would like to recast
> my question. I hope this is not inappropriate.
>
> I have data in panel format and I declared it as a a panel. All fine.
> I am carrying out a fixed effect regression to overcome the
> endogeneity problem (due to simultaneity) I have in my model. In fact,
> to demonstrate that I have the endogeneity problem, I wanted to
> compare the estimates from a plain OLS regression (regress y x) to the
> estimates from the fixed effects regression (xtreg y x, fe). To carry
> out the OLS regression, I simply run the command "regress y x".
>
> My question is: If I use the plain "regress y x" command to get the
> plain OLS estimates using the panel data I have, will that regression
> be correct? Will Stata treat my panel data as pooled cross sections?
> Or do I need to instruct Stata with a special command to run the plain
> OLS using my panel data?
*
* 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/