two way or one way fixed effect model

[Noor Hend]

Hello,

I am writing my master thesis on the effect of implementing GSCM practices on financial performance. I am using a balanced panel data set across 9 year periods with 464 annual observations (i.e. firms). By performing a Hausman test, it says that random effect model is more suitable (p=0,0735)
However, I am not sure if I should include dummy variables for both year and ID, as now only ID dummies are considered. So how do I examine whether to use a one way or two way model in general (i.e. including both ID and year, or only one of the two). And when using a one way model, how to I determine whether to include ID dummies or year dummies. This decision probably has to be made before doing the Hausman test I think?

 xtset ID year, yearly
       panel variable:  ID (strongly balanced)
        time variable:  year, 2006 to 2014
                delta:  1 year

xtreg TobinsQ laggedGSCMP Firmrisk Firmsize Industry, fe
note: Industry omitted because of collinearity

Fixed-effects (within) regression               Number of obs      =      3712
Group variable: ID                              Number of groups   =       464

R-sq:  within  = 0.0050                         Obs per group: min =         8
       between = 0.0356                                        avg =       8.0
       overall = 0.0247                                        max =         8

                                                F(3,3245)          =      5.49
corr(u_i, Xb)  = -0.0538                        Prob > F           =    0.0009

------------------------------------------------------------------------------
     TobinsQ |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
 laggedGSCMP |  -.1490492    .108578    -1.37   0.170    -.3619376    .0638393
    Firmrisk |  -.2108832   .1473064    -1.43   0.152    -.4997062    .0779398
    Firmsize |  -.1545399   .0489047    -3.16   0.002    -.2504271   -.0586528
    Industry |          0  (omitted)
       _cons |   3.295451   .3674794     8.97   0.000     2.574936    4.015967
-------------+----------------------------------------------------------------
     sigma_u |  1.0374074
     sigma_e |  .83317079
         rho |  .60789743   (fraction of variance due to u_i)
------------------------------------------------------------------------------
F test that all u_i=0:     F(463, 3245) =    12.22           Prob > F = 0.0000

. estimates store fixed

. xtreg TobinsQ laggedGSCMP Firmrisk Firmsize Industry, re

Random-effects GLS regression                   Number of obs      =      3712
Group variable: ID                              Number of groups   =       464

R-sq:  within  = 0.0049                         Obs per group: min =         8
       between = 0.0604                                        avg =       8.0
       overall = 0.0407                                        max =         8

                                                Wald chi2(4)       =     45.61
corr(u_i, X)   = 0 (assumed)                    Prob > chi2        =    0.0000

------------------------------------------------------------------------------
     TobinsQ |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
 laggedGSCMP |   -.071065   .1041549    -0.68   0.495    -.2752048    .1330748
    Firmrisk |  -.1877267   .1332181    -1.41   0.159    -.4488294     .073376
    Firmsize |  -.1471782    .027809    -5.29   0.000    -.2016829   -.0926735
    Industry |  -.0929571   .0275097    -3.38   0.001    -.1468751   -.0390391
       _cons |   3.643309   .2489337    14.64   0.000     3.155407     4.13121
-------------+----------------------------------------------------------------
     sigma_u |  .97660797
     sigma_e |  .83317079
         rho |    .578762   (fraction of variance due to u_i)
------------------------------------------------------------------------------

. estimates store random

. hausman fixed random

                 ---- Coefficients ----
             |      (b)          (B)            (b-B)     sqrt(diag(V_b-V_B))
             |     fixed        random       Difference          S.E.
-------------+----------------------------------------------------------------
 laggedGSCMP |   -.1490492     -.071065       -.0779841        .0306749
    Firmrisk |   -.2108832    -.1877267       -.0231565        .0628659
    Firmsize |   -.1545399    -.1471782       -.0073617        .0402284
------------------------------------------------------------------------------
                           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(3) = (b-B)'[(V_b-V_B)^(-1)](b-B)
                          =        6.95
                Prob>chi2 =      0.0735

[Phil Bromiley]

You'll increase your chances of a useful answer by following the FAQ on asking questions - provide Stata code in code delimiters, readable Stata output, and sample data using dataex.

You don't include ID as a dummy - the xtreg takes care of that. You enter year as i.year. You can then test if the year dummies are jointly zero (after xtreg, run xtreg,coefl to see how to refer to the dummies).