invariant regressors

[Vaggelis Ktenas]

Dear all,

I can't understand what's going on.. I have panel data and when i run BE in stata, individual-invariant regressor is not dropping out. I actually want to use a FE model with the individual-invariant variable and i used between estimator to check the result.. My individual-invariant regressor is tax rate which is the same for each company (and changes every one or three years). I have only one regressor (taxrate) and my depvariable and i haven't add time variable in model due to collinearity with rate.

Thank you

[Carlo Lazzaro]

Vaggelis:
the trivial explanation is that, unlike within estimator, -be- estimator is not supposed to wipe-out time-invariant predictors, as you can see from the following toy-example:

. use "http://www.stata-press.com/data/r15/nlswork.dta"
(National Longitudinal Survey.  Young Women 14-26 years of age in 1968)

. xtreg ln_wage i.race, be

Between regression (regression on group means)  Number of obs     =     28,534
Group variable: idcode                          Number of groups  =      4,711

R-sq:                                           Obs per group:
     within  = 0.0000                                         min =          1
     between = 0.0198                                         avg =        6.1
     overall = 0.0186                                         max =         15

                                                F(2,4708)         =      47.53
sd(u_i + avg(e_i.))=  .4204359                  Prob > F          =     0.0000

------------------------------------------------------------------------------
     ln_wage |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
        race |
      black  |  -.1280888   .0136568    -9.38   0.000    -.1548625   -.1013151
      other  |   .1127399   .0561628     2.01   0.045     .0026345    .2228453
             |
       _cons |   1.685266   .0072869   231.27   0.000     1.670981    1.699552
------------------------------------------------------------------------------

As an aside, I fail to get why checking the results of a -fe- panel data regression with a pretty different specification, such as -be-.

[Phil Bromiley]

You should look at user-written xthybrid and Mundlak estimators.