xtabond2 - period variables dropped due to collinearity

Statistische Auswertung von Longitudinal- und Panel-Daten mit Stata.

xtabond2 - period variables dropped due to collinearity

Beitragvon maiss » Do 19. Jun 2014, 18:01

Hallo allesamt,

ich arbeite mit einem unbalanced panel über sechs Zeitperioden (0-5).
Ich möchte eine GMM-Regression mit xtabond2 machen, klappt auch alles soweit, jedoch kommt immer die Meldung:

0b.period dropped due to collinearity
5.period dropped due to collinearity

Kann mir das vielleicht jemand bitte erklären?

Code: Alles auswählen
. xtabond2  FDI
> DR2
> DR Ext_Debt ODA GDP_growth log_GDP_pc
> Inv_Total Inv_Edu Inv_Health
> Inflation Conflicts School PRCL i.period,
> twostep robust small
> gmmstyle(DR2 DR Ext_Debt log_GDP_pc Conflicts School PRCL
> L.(ODA GDP_growth Inv_Total Inv_Edu Inv_Health Inflation),collapse)
> ivstyle(i.period);
Favoring space over speed. To switch, type or click on mata: mata set matafavor
>  speed, perm.
0b.period dropped due to collinearity
5.period dropped due to collinearity
Warning: Two-step estimated covariance matrix of moments is singular.
  Using a generalized inverse to calculate optimal weighting matrix for two-ste
> p estimation.
  Difference-in-Sargan/Hansen statistics may be negative.

Dynamic panel-data estimation, two-step system GMM
------------------------------------------------------------------------------
Group variable: country                         Number of obs      =       358
Time variable : period                          Number of groups   =       103
Number of instruments = 76                      Obs per group: min =         1
F(17, 102)    =    318.91                                      avg =      3.48
Prob > F      =     0.000                                      max =         5
------------------------------------------------------------------------------
             |              Corrected
         FDI |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
         DR2 |   154.9574   23.40747     6.62   0.000     108.5288     201.386
          DR |   -34.0646   12.76118    -2.67   0.009    -59.37634   -8.752868
    Ext_Debt |   .6316289   2.049977     0.31   0.759    -3.434491    4.697749
         ODA |   18.07086   13.83563     1.31   0.194    -9.372046    45.51376
  GDP_growth |  -.0790777   .1638505    -0.48   0.630    -.4040744     .245919
  log_GDP_pc |   .9881322   1.067983     0.93   0.357    -1.130206    3.106471
   Inv_Total |   .3233997   .1116049     2.90   0.005      .102032    .5447674
     Inv_Edu |  -.7230317   .5421905    -1.33   0.185    -1.798464    .3524006
  Inv_Health |  -.1340951   .6771394    -0.20   0.843    -1.477198    1.209008
   Inflation |   .3185919   .7958662     0.40   0.690    -1.260005    1.897189
   Conflicts |  -.1616975   1.509193    -0.11   0.915    -3.155175     2.83178
      School |   -.043785   .0458947    -0.95   0.342     -.134817    .0472469
        PRCL |    .065045   .6166273     0.11   0.916    -1.158032    1.288122
             |
      period |
          1  |  -2.476416   1.328475    -1.86   0.065     -5.11144    .1586083
          2  |  -.4240452   .9174682    -0.46   0.645    -2.243839    1.395749
          3  |   .1601625    .956057     0.17   0.867    -1.736172    2.056497
          4  |   .9107486   .9487632     0.96   0.339    -.9711187    2.792616
             |
       _cons |  -4.031076   11.61351    -0.35   0.729    -27.06642    19.00427
------------------------------------------------------------------------------
Instruments for first differences equation
  Standard
    D.(0b.period 1.period 2.period 3.period 4.period 5.period)
  GMM-type (missing=0, separate instruments for each period unless collapsed)
    L(1/5).(DR2 DR Ext_Debt log_GDP_pc Conflicts School PRCL L.ODA
    L.GDP_growth L.Inv_Total L.Inv_Edu L.Inv_Health L.Inflation) collapsed
Instruments for levels equation
  Standard
    0b.period 1.period 2.period 3.period 4.period 5.period
    _cons
  GMM-type (missing=0, separate instruments for each period unless collapsed)
    D.(DR2 DR Ext_Debt log_GDP_pc Conflicts School PRCL L.ODA L.GDP_growth
    L.Inv_Total L.Inv_Edu L.Inv_Health L.Inflation) collapsed
maiss
 
Beiträge: 3
Registriert: Fr 30. Mai 2014, 12:33
Danke gegeben: 0
Danke bekommen: 0 mal in 0 Post

Re: xtabond2 - period variables dropped due to collinearity

Beitragvon mangel76 » Fr 25. Jul 2014, 12:44

Ist zwar schon was älter, aber vielleicht ist es ja noch interessant.

Bei System GMM, was du hier ja schätzt, fällt mindestens die erste Periode weg, da hier die Level mit den ersten Differenzen instrumentiert werden. In der ersten Periode gibt es keine ersten Differenzen, im verwendbaren Datensatz (Perioden 1-5) sind also alle P0-Dummies gleich 0. Jetzt hast du für die verbleibenden 5 Perioden (1-5) 5 Zeitdummys, also schmeißt Stata noch eine raus, in diesem Fall die letzte.

Viele Grüße
Mangel
mangel76
 
Beiträge: 39
Registriert: Do 9. Jan 2014, 16:44
Danke gegeben: 0
Danke bekommen: 0 mal in 0 Post


Zurück zu Longitudianal und Panel-Analyse

Wer ist online?

Mitglieder in diesem Forum: 0 Mitglieder und 3 Gäste

cron