st: specify base year for interacted factor variable.

[Daifeng He <dhe.statlist@gmail.com>]

Dear all,

I am running a panel model regressing y on a continous variable x,
year dummies, and an interaction of x with the year dummies.  I have
three years of data 1999, 2000 and 2001 and I wan to specify the first
year 1999 to be the base year.  However, when trying to use the factor
variable commands to do the job, I can only sucessfully make the 1999
as baseyear for the year dummies, but I cannot make 1999 as the base
year for the interaction term -- it's always 2001 which is dropped.
(Please see the Stata output below).

Does anyone know how I can get 1999 to be the base year for the
year#c.x interaction term?   Any insight would be appreciated!


Thanks,
Daifeng


. xtreg y   ib1999.year#c.x  i.year,  i(id) fe
note: 2001.year#c.x omitted because of collinearity

Fixed-effects (within) regression               Number of obs      =      3027
Group variable: id                              Number of groups   =      1029

R-sq:  within  = 0.0036                         Obs per group: min =         1
       between = 0.0076                                        avg =       2.9
       overall = 0.0046                                        max =         3

                                                F(4,1994)          =      1.81
corr(u_i, Xb)  = 0.0409                         Prob > F           =    0.1252

------------------------------------------------------------------------------
           y |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
    year#c.x |
       1999  |   3.345423   3.330076     1.00   0.315     -3.18537    9.876217
       2000  |   1.934874   3.289485     0.59   0.556    -4.516315    8.386062
       2001  |  (omitted)
             |
        year |
       2000  |   1.422862   1.469454     0.97   0.333    -1.458965    4.304689
       2001  |   3.554589    1.46647     2.42   0.015     .6786138    6.430563
             |
       _cons |   195.3646    1.04039   187.78   0.000     193.3243     197.405
-------------+----------------------------------------------------------------
     sigma_u |  55.369127
     sigma_e |   32.73169
         rho |  .74103537   (fraction of variance due to u_i)
------------------------------------------------------------------------------
F test that all u_i=0:     F(1028, 1994) =     6.85          Prob > F = 0.0000
*
*   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/