st: RE: Question on controlling for associated observations

[Jan Bryla <JBR@finansraadet.dk>]

Dear Elena

If I understand you correctly you still need to identify which two counties should have common fixed-effects. Once this is done (which you could still do the way I described in the previous post), I assume you could do the following:

- generate dummy variables for each pair and then perform fixed-effects using LSDV (xi: reg Y X1... Xk i.pair)
- generate a group variable for the chosen pair, then xtsetting 'group' time and performing -xtreg, re-

Thus, I guess you are right in that this should reduce the number of parameters (unless you are using LSDV with a large N and small t)

Let me know if this comes closer to solving your problem.

Best,
Jan 



Jan Bryla
The Danish Bankers Association



-----Original Message-----
From: owner-statalist@hsphsun2.harvard.edu [mailto:owner-statalist@hsphsun2.harvard.edu] On Behalf Of Elena Quercioli
Sent: 12. marts 2011 23:44
To: statalist@hsphsun2.harvard.edu
Subject: st: Question on controlling for associated observations

Dear Statalist-ers,

I need to formulate in STATA a panel regression in which I tie down
pairs of observations to have the same fixed effects.  In my case  (I
am using a random effects MLE estimation procedure) I have a set of
data for counties along time zone lines. I want to
estimate the effect on the observations of landing on the west side of
the time zone line. Of course, I also have other controls. But I want
to think of the counties that are adjacent as pairs (clusters of two),
and thus estimating a 'common fixed effects' through the years.

To be clear my panel regression is:

y_itk = a_0 + a_1*dummy1_ik + a_2*dummy2_ik + a_3*dummy3_i + epsilon_i
+ epsilon_t + epsilon_it + epsilon_itk

Where i is the county identifier, t is years in panel and k is west.

I am testing the significance of the coefficient a_1 of west (dummy1_ik=1)

Does anyone know how to do this? Thank you in advance for your help.

Elena

*
*   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/

*
*   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/