st: RE: Panel data: large number of linear time trends

[Nick Cox <n.j.cox@durham.ac.uk>]

Maligning all those who presumably gave advice on this with a blanket criticism is not the best way to win friends and influence people, to allude to a once well-known book. 

Why is this thought or implied to be so difficult? 

. webuse grunfeld, clear

. statsby slope=_b[year] , by(company) : regress mvalue year
(running regress on estimation sample)

      command:  regress mvalue year
        slope:  _b[year]
           by:  company

Statsby groups
----+--- 1 ---+--- 2 ---+--- 3 ---+--- 4 ---+--- 5 
..........

. l

     +--------------------+
     | company      slope |
     |--------------------|
  1. |       1   56.35872 |
  2. |       2    1.46263 |
  3. |       3   7.183384 |
  4. |       4    8.91579 |
  5. |       5   10.77564 |
     |--------------------|
  6. |       6   33.69406 |
  7. |       7   .4736842 |
  8. |       8    27.8215 |
  9. |       9   7.434586 |
 10. |      10   -.563985 |
     +--------------------+

Nick 
n.j.cox@durham.ac.uk 

William Gui Woolston

I am estimating a panel data model, where the unit of observation is a
county-year.  There are roughly 3,100 counties in the United States,
and I have data for 12 years.

I wish to include linear county-time trends.  That is, I want a
separate time trend for each county.

Estimating this model by "brute force" (by interacting time with a
dummy for each county) would mean having an additional 3,100 variables
to my model.  Is there a more efficient way to estimate this model?

Thank you so much for your consideration.

William

PS.  Note that some versions of this question have appeared in earlier
Statalist threads
(http://www.stata.com/statalist/archive/2011-05/msg00035.html, for
exacmple), but none of them provides a satisfactory answer.

*
*   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/