Announcement

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

  • order of exogenous variables in sfpanel model (stochastic frontier model)

    Dear, all

    I have some questions during my running STATA. I am running a stochastic frontier approach, based on Battese and Coelli's (1995) model, by using sfpanel code(developed by Belotti et al., (2013)).
    I have 6 independent variables, time trend (t), and 4 regional dummy variables in the stochastic frontier. As for the inefficiency function, I used 4 explanatory variables and 2 dummy variables.
    I made data scaled and cleaned, so that I could obtain the result as followed:

    Code:
     sfpanel $output $labor $land $fert $capital $climate_0 $dum t $dummy, model(bc95) d(t) emean(export_fix urb_ratio mean_sch gdp_corr flo_dum1 drought_dum1)
    Code:
    Inefficiency effects model (truncated-normal)        Number of obs =       609
    Group variable: id                                Number of groups =        31
    Time variable: year                             Obs per group: min =        13
                                                                   avg =      19.6
                                                                   max =        20
    
                                                         Prob > chi2   =    0.0000
    Log likelihood =  -155.2237                          Wald chi2(11)  =   5580.29
    
    ------------------------------------------------------------------------------
     lnvalue_agr |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
    -------------+----------------------------------------------------------------
    Frontier     |
         lnlabor |   .6297881   .0248619    25.33   0.000     .5810598    .6785165
          lnland |   .1455606   .0340051     4.28   0.000     .0789117    .2122094
          lnfert |   .0865237   .0089076     9.71   0.000     .0690651    .1039823
       lncapital |   .1419817   .0183148     7.75   0.000     .1060855     .177878
            temp |  -.0017632   .0097441    -0.18   0.856    -.0208614    .0173349
            prcp |   .0005432   .0000461    11.79   0.000      .000453    .0006335
            dum2 |   .1528198   .0635501     2.40   0.016     .0282638    .2773758
            dum3 |   1.893081   .0909385    20.82   0.000     1.714845    2.071318
            dum4 |   1.850183   .1083842    17.07   0.000     1.637754    2.062612
            dum5 |   .4453163   .0715977     6.22   0.000     .3049874    .5856451
               t |   .0169506   .0023294     7.28   0.000      .012385    .0215162
           _cons |   2.837632   .3713127     7.64   0.000     2.109872    3.565391
    -------------+----------------------------------------------------------------
    Mu           |
      export_fix |  -2.636783   .4892806    -5.39   0.000    -3.595755   -1.677811
       urb_ratio |  -3.725086   .5167533    -7.21   0.000    -4.737904   -2.712268
        mean_sch |   .4050751   .0432029     9.38   0.000      .320399    .4897512
        gdp_corr |   .3778436   .0723724     5.22   0.000     .2359963     .519691
        flo_dum1 |   .0371779   .0536684     0.69   0.488    -.0680103    .1423661
    drought_dum1 |   .0419306   .0659389     0.64   0.525    -.0873074    .1711686
           _cons |  -1.005427   .1505207    -6.68   0.000    -1.300442   -.7104116
    -------------+----------------------------------------------------------------
    Usigma       |
           _cons |  -5.386039    1.25924    -4.28   0.000    -7.854104   -2.917974
    -------------+----------------------------------------------------------------
    Vsigma       |
           _cons |  -2.340532   .0602993   -38.82   0.000    -2.458716   -2.222347
    -------------+----------------------------------------------------------------
         sigma_u |   .0676763   .0426103     1.59   0.112     .0197017    .2324717
         sigma_v |   .3102844    .009355    33.17   0.000     .2924802    .3291724
          lambda |   .2181105   .0459056     4.75   0.000     .1281372    .3080838
    ------------------------------------------------------------------------------
    However, something interesting is that when I run the code, by changing the order of exogenous variables (see below the command "emean"), the result comes out differently with NAs.
    Here is one of the examples as followed:

    Code:
     sfpanel $output $labor $land $fert $capital $climate_0 $dum t $dummy, model(bc95) d(t) emean(urb_ratio export_fix mean_sch gdp_corr flo_dum1 drought_dum1)
    Code:
    Inefficiency effects model (truncated-normal)        Number of obs =       609
    Group variable: id                                Number of groups =        31
    Time variable: year                             Obs per group: min =        13
                                                                   avg =      19.6
                                                                   max =        20
    
                                                         Prob > chi2   =    0.0000
    Log likelihood =  -170.1489                          Wald chi2(11)  =   5347.07
    
    ------------------------------------------------------------------------------
     lnvalue_agr |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
    -------------+----------------------------------------------------------------
    Frontier     |
         lnlabor |   .6136596   .0253306    24.23   0.000     .5640125    .6633067
          lnland |   .1913424   .0362745     5.27   0.000     .1202457    .2624391
          lnfert |   .0979961   .0092775    10.56   0.000     .0798126    .1161796
       lncapital |   .1288264    .019428     6.63   0.000     .0907482    .1669046
            temp |   -.014941   .0101247    -1.48   0.140    -.0347851    .0049032
            prcp |   .0006067   .0000501    12.11   0.000     .0005086    .0007049
            dum2 |   .1929125   .0638062     3.02   0.002     .0678545    .3179704
            dum3 |   1.885857    .097887    19.27   0.000     1.694002    2.077712
            dum4 |   1.689854   .1175275    14.38   0.000     1.459504    1.920204
            dum5 |   .4981003   .0717566     6.94   0.000       .35746    .6387406
               t |   .0139226   .0024617     5.66   0.000     .0090978    .0187474
           _cons |   2.968205   .3810191     7.79   0.000     2.221422    3.714989
    -------------+----------------------------------------------------------------
    Mu           |
       urb_ratio |  -3.521951   .6684924    -5.27   0.000    -4.832172    -2.21173
      export_fix |  -3.473571          .        .       .            .           .
        mean_sch |   .4193095   .0521165     8.05   0.000     .3171631     .521456
        gdp_corr |   .3108142     .10723     2.90   0.004     .1006473    .5209811
        flo_dum1 |   .0693675   .0751694     0.92   0.356    -.0779617    .2166968
    drought_dum1 |   .0505226   .0909306     0.56   0.578    -.1276981    .2287433
           _cons |  -1.111038   .2249489    -4.94   0.000    -1.551929   -.6701459
    -------------+----------------------------------------------------------------
    Usigma       |
           _cons |  -3.888564          .        .       .            .           .
    -------------+----------------------------------------------------------------
    Vsigma       |
           _cons |  -2.294794   .0645472   -35.55   0.000    -2.421305   -2.168284
    -------------+----------------------------------------------------------------
         sigma_u |   .1430899          .        .       .            .           .
         sigma_v |    .317462   .0102456    30.99   0.000     .2980028    .3381918
          lambda |   .4507308          .        .       .            .           .
    ------------------------------------------------------------------------------
    I wonder how the results are different if I changed the order of variables.

    Is there something that I've missed? or something wrong?
    Last edited by Sharon Bang; 21 Dec 2021, 00:29.

  • #2
    Does anyone know or faced this kind of problem? If you give me the comments, it would be really helpful to me!

    Comment

    Working...
    X