Bookmark and Share

Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

st: Panel analysis - Chosing the best estimator


From   "Stefan Weih" <[email protected]>
To   [email protected]
Subject   st: Panel analysis - Chosing the best estimator
Date   Wed, 2 May 2012 11:51:58 +0200

Dear Statlist-members,

I am currently analysing some panel data and I am still in the process of
finding the best estimator for my data.
I ran several tests (e.g. hausman) already (please see below) which
indicate that fixed-effects (FE) would be the right one. However, the data
includes also some time-invariant variables (e.g. age, dummies for country
of origin, dummies for certain sequence effect)that I would like to
estimate since they are of high importance for my research topic.
So, if FE was the right estimator to use, from my understanding, I do not
get any coefficients on my time-invariant variables.

Do you have any recommendations for me from your experience what to do in
such a situation? What other estimators or models would you recommend? And
what could be a possible line of argumentation to use this estimator/model
then?

Thank you very much for any help!

Best regards,
Stefan


Here are the test results on the hausman test from the base model (some
additional info: both heteroscedasticit and autocorrelation is present):

quietly xtreg _roa _entropy_p_4dig_gics _entropy_i _firm_size_rev_nat_log
firm_age _leverage_assets _rd_intensity coo_eu rope coo_asia 
__ext_growth_view2 roa_industry_sic industrygrowth_sic
industry_2dig_gics_10 industry_2dig_gics_15 industry_2dig_gics_20
industry_2dig_gics_25 industry_2dig_gics_30 industry_2dig_gics_35
industry_2dig_gics_50 industry_2dig_gics_55 year_2001 year _2002 year_2003
year_2004 year_2005 year_2006 year_2007 year_2008 year_2009 year_2010, re
vce(robust)
estimates store Model1_RE
xttest0

Breusch and Pagan Lagrangian multiplier test for random effects

        _roa[_geo_id,t] = Xb + u[_geo_id] + e[_geo_id,t]

        Estimated results:
                         |       Var     sd = sqrt(Var)
                ---------+-----------------------------
                    _roa |   .0092097       .0959672
                       e |   .0020787       .0455924
                       u |   .0057581       .0758819

        Test:   Var(u) = 0
                             chibar2(01) =  1784.49
                          Prob > chibar2 =   0.0000

.
. testparm year_*

 ( 1)  year_2001 = 0
 ( 2)  year_2002 = 0
 ( 3)  year_2003 = 0
 ( 4)  year_2004 = 0
 ( 5)  year_2005 = 0
 ( 6)  year_2006 = 0
 ( 7)  year_2007 = 0
 ( 8)  year_2008 = 0
 ( 9)  year_2009 = 0
 (10)  year_2010 = 0

           chi2( 10) =   39.43
         Prob > chi2 =    0.0000

quietly xtreg _roa _entropy_p_4dig_gics _entropy_i _firm_size_rev_nat_log
firm_age _leverage_assets _rd_intensity coo_eu rope coo_asia 
__ext_growth_view2 roa_industry_sic industrygrowth_sic
industry_2dig_gics_10 industry_2dig_gics_15 industry_2dig_gics_20
industry_2dig_gics_25 industry_2dig_gics_30 industry_2dig_gics_35
industry_2dig_gics_50 industry_2dig_gics_55 year_2001 year _2002 year_2003
year_2004 year_2005 year_2006 year_2007 year_2008 year_2009 year_2010, fe
vce(robust)
estimates store Model2_FE

hausman Model1_RE Model2_FE, sigmamore


                       ---- Coefficients ----
                   |      (b)          (B)            (b-B)    
sqrt(diag(V_b-V_B))
                   |    FE_Base      RE_Base       Difference          S.E.
      -------------+----------------------------------------------------------------
      _entropy_p~s |   -.0065339    -.0128126        .0062786        .0039785
        _entropy_i |   -.0322273    -.0149911       -.0172362        .0071288
      _firm_size~g |   -.0063533    -.0044396       -.0019136        .0035541
      _leverage~ts |   -.0673979    -.0772317        .0098338        .0054994
      _rd_intens~y |   -1.356907    -1.014656       -.3422504        .0320651
      __ext_grow~2 |    .0103768     .0110762       -.0006994        .0007249
      roa_indust~c |    .1121113     .1193533       -.0072419        .0066685
      industrygr~c |    .0003156     .0002818        .0000337        .0000503
         year_2001 |   -.0249425     -.029257        .0043145        .0008362
         year_2002 |   -.0257642    -.0305598        .0047956        .0008249
         year_2003 |   -.0095334    -.0155798        .0060464        .0008552
         year_2004 |   -.0000192    -.0065239        .0065047        .0010287
         year_2005 |   -.0002363    -.0047472         .004511        .0011081
         year_2006 |   -.0018331    -.0069484        .0051153         .001359
         year_2007 |     .002791    -.0028696        .0056606        .0017221
         year_2008 |    .0044858    -.0010075        .0054933        .0020172
         year_2009 |   -.0021243    -.0079116        .0057873        .0017457
         year_2010 |    .0063719      .000954        .0054179        .0019359
      ------------------------------------------------------------------------------
                                 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(18) = (b-B)'[(V_b-V_B)^(-1)](b-B)
                                =      137.80
                      Prob>chi2 =      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/


© Copyright 1996–2018 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   Site index