For fixed-effect -xtreg, fe- for random effect -xtreg, re-. I think xtgee
computes random-effects. Probably you would like to take a look of
Hausman-Taylor estimator as well -help xthtaylor-. Rodrigo.
----- Original Message -----
From: "Christer Thrane" <[email protected]>
To: <[email protected]>
Sent: Saturday, May 27, 2006 2:19 PM
Subject: Re: st: Re: Time invariant variables and panel data
First q: yes
Second q: No, I don't think so. FE utilizes the changes within persons, so
if thing don't change, FE is not viable.
CT
----- Original Message -----
From: "Rho YH" <[email protected]>
To: <[email protected]>
Sent: Saturday, May 27, 2006 7:44 PM
Subject: RE: st: Re: Time invariant variables and panel data
> To be sure, are the "random effects" and "fixed effects" of the answer
> the -RE and -FE options of xt-
> series commands?
> Also, as xtgee does not have such option, then can I use time invariant
> variables anyway?
>> ---- Original Message ----
>> From : Christer Thrane [[email protected]]
>> To : [email protected]
>> Date : 2006? 5? 28?(?) 01:38:38
>> Subject : st: Re: Time invariant variables and panel data
>>
>>
>>Hi,
>>
>>If you estimate random effects models, you can have both time-varying and
>>time-invariant variables on the right hand side.
>>
>>If you estimate fixed effects models, you must have time-varying on the
>>right hand side. But fixed effects also handles interactions between
>>time-varying and time-invariant right hand side variables. (Then the
>>interaction is no longer time-invariant.)
>>
>>Take a look at Woldridge's Basic Econometrics. A modern introduction--it's
>>on Stata website.
>>
>>HTH
>>
>>Christer
>>************************************************
>>Professor Christer Thrane
>>Department of Social Science
>>Lillehammer University College
>>2626 Lillehammer, Norway
>>+ 47 61 28 81 70 (fax)
>>+ 47 61 28 82 47 (phone, work)
>>+ 47 61 25 53 04 (phone, home)
>>E-mail, work: [email protected]
>>E-mail, home: [email protected]
>>************************************************
>>
>>
>>
>>----- Original Message -----
>>From: "Rho YH"
>>To:
>>Sent: Saturday, May 27, 2006 4:48 PM
>>Subject: st: Time invariant variables and panel data
>>
>>
>>>I have incidentally found that panel data models (as far as I found the
>>>Statalist, xtreg with -FE option)
>>> cannot have time invariant variables and will be automatically omitted
>>> when included.
>>> In a general sense, then all xt- models (including xtgee) cannot have
>>> time
>>> invariant variables?
>>> Or in what instance does this fact come in effect?
>>> Second, then does this mean that I cannot use time invariant variables
>>> (like sex (M/F)) even
>>> when these variables may be important?
>>> I have two books about panel data (Twisk, and Sophia Rabe-Hesketh) but
>>> did
>>> not find this fact.
>>> Is it written in these books?
*
* 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/
