Announcement

Collapse
No announcement yet.
X
  • Filter
  • Time
  • Show
Clear All
new posts

  • Fixed Effect - Testing for Time fixed effects

    Dear all,

    I'm having a Problem when testing for time fixed effects. My panel data includes three years:



    sum year

    Variable | Obs Mean Std. Dev. Min Max
    -------------+---------------------------------------------------------
    year | 1,221 2015 .8168311 2014 2016


    I did the hausman test do test whether to use fixed or random effects:


    . hausman fixed random

    ---- Coefficients ----
    | (b) (B) (b-B) sqrt(diag(V_b-V_B))
    | fixed random Difference S.E.
    -------------+----------------------------------------------------------------
    shareholder | .0612042 -.1386216 .1998258 .0460368
    partnership | -5.862131 -20.46098 14.59885 5.187308
    ------------------------------------------------------------------------------
    b = consistent under Ho and Ha; obtained from xtreg
    B = inconsistent under Ha, efficient under Ho; obtained from xtreg

    Test: Ho: difference in coefficients not systematic

    chi2(2) = (b-B)'[(V_b-V_B)^(-1)](b-B)
    = 27.71
    Prob>chi2 = 0.0000


    After that I wanted to check for time fixed effects using:

    xtreg $ylist $xlist i.year, fe
    testparm i.year

    However, in the results 2014 is missing:


    . xtreg $ylist $xlist i.year, fe

    Fixed-effects (within) regression Number of obs = 776
    Group variable: id Number of groups = 401

    R-sq: Obs per group:
    within = 0.0170 min = 1
    between = 0.3403 avg = 1.9
    overall = 0.1771 max = 3

    F(4,371) = 1.61
    corr(u_i, Xb) = 0.3769 Prob > F = 0.1722

    ------------------------------------------------------------------------------
    esgscore | Coef. Std. Err. t P>|t| [95% Conf. Interval]
    -------------+----------------------------------------------------------------
    shareholder | .0209226 .0558744 0.37 0.708 -.0889477 .1307929
    partnership | -4.89243 5.504378 -0.89 0.375 -15.71612 5.931263
    |
    year |
    2015 | 3.054609 1.9083 1.60 0.110 -.6978307 6.807049
    2016 | 3.913033 1.956296 2.00 0.046 .0662144 7.759851
    |
    _cons | 45.99972 2.780761 16.54 0.000 40.53169 51.46775
    -------------+----------------------------------------------------------------
    sigma_u | 14.991157
    sigma_e | 16.064203
    rho | .46548851 (fraction of variance due to u_i)
    ------------------------------------------------------------------------------
    F test that all u_i=0: F(400, 371) = 1.68 Prob > F = 0.0000

    . testparm i.year

    ( 1) 2015.year = 0
    ( 2) 2016.year = 0

    F( 2, 371) = 2.04
    Prob > F = 0.1318


    Anybody with an idea where the mistake might be? Thank you in advance!

  • #2
    Konstantin:
    welocme to this forum.
    No mistake there: Stata simply omits one year to avoid the so called dummy trap (https://en.wikipedia.org/wiki/Dummy_...le_(statistics)).
    Kind regards,
    Carlo
    (StataNow 18.5)

    Comment


    • #3
      Thank you for the quick reply Carlo, I'm relieved

      Comment

      Working...
      X