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]

Re: st: SFE technical estimates STAT and SAS


From   Scott Merryman <[email protected]>
To   [email protected]
Subject   Re: st: SFE technical estimates STAT and SAS
Date   Wed, 9 May 2012 15:41:01 -0500

Yes.  Using the greene9 data set:

. webuse greene9,clear

. frontier lnv lnk lnl, nolog

Stoc. frontier normal/half-normal model           Number of obs   =         25
                                                 Wald chi2(2)    =     743.71
Log likelihood =  2.4695222                       Prob > chi2     =     0.0000

------------------------------------------------------------------------------
        lnv |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
        lnk |   .2585478    .098764     2.62   0.009     .0649738    .4521218
        lnl |   .7802451   .1199399     6.51   0.000     .5451672    1.015323
      _cons |   2.081135    .281641     7.39   0.000     1.529128    2.633141
-------------+----------------------------------------------------------------
   /lnsig2v |   -3.48401   .6195353    -5.62   0.000    -4.698277   -2.269743
   /lnsig2u |  -3.014599    1.11694    -2.70   0.007    -5.203761   -.8254368
-------------+----------------------------------------------------------------
    sigma_v |   .1751688   .0542616                      .0954514    .3214633
    sigma_u |   .2215073   .1237052                       .074134    .6618486
     sigma2 |   .0797496   .0426989                     -.0039388     .163438
     lambda |   1.264536   .1678684                      .9355204    1.593552
------------------------------------------------------------------------------
Likelihood-ratio test of sigma_u=0: chibar2(01) = 0.43   Prob>=chibar2 = 0.256

. predict  te, te

. cl state te

                    state         te
  1.              Alabama   .8231754
  2.           California   .8692654
  3.          Connecticut   .8318406
  4.              Florida   .6016339
  5.              Georgia   .9040509
  6.             Illinois   .8891712
  7.              Indiana   .8150898
  8.                 Iowa    .785352
  9.               Kansas   .9066431
 10.             Kentucky   .9464481
 11.            Louisiana   .8214222
 12.                Maine   .8061242
 13.             Maryland   .8772239
 14.        Massachusetts   .8596337
 15.             Michigan   .8582037
 16.             Missouri   .9050045
 17.            NewJersey   .9111356
 18.              NewYork   .7636376
 19.                 Ohio   .8010177
 20.         Pennsylvania   .8648741
 21.                Texas   .8217006
 22.             Virginia   .8732988
 23.           Washington    .898432
 24.         WestVirginia   .8602608
 25.            Wisconsin   .8727364

 From SAS:
 proc qlim data=greene  covest=hessian;
 model lnv =  lnk lnl;
 endogenous lnv ~ frontier (type=HALF production);
 output out = work_out1 TE1  ;
 run;
 proc print data = work_out1;
 var state te1;
 run;

                            Parameter Estimates

                                          Standard                 Approx
   Parameter    DF        Estimate           Error    t Value    Pr > |t|

   Intercept     1        2.081135        0.281648       7.39     <.0001
   lnk           1        0.258548        0.098766       2.62     0.0089
   lnl           1        0.780245        0.119945       6.51     <.0001
   _Sigma_v      1        0.175169        0.054265       3.23     0.0012
   _Sigma_u      1        0.221507        0.123706       1.79     0.0734

       Obs    state                     TE1

         1    Alabama                 0.82318
         2    California              0.86927
         3    Connecticut             0.83184
         4    Florida                 0.60163
         5    Georgia                 0.90405
         6    Illinois                0.88917
         7    Indiana                 0.81509
         8    Iowa                    0.78535
         9    Kansas                  0.90664
        10    Kentucky                0.94645
        11    Louisiana               0.82142
        12    Maine                   0.80612
        13    Maryland                0.87722
        14    Massachusetts           0.85963
        15    Michigan                0.85820
        16    Missouri                0.90500
        17    NewJersey               0.91114
        18    NewYork                 0.76364
        19    Ohio                    0.80102
        20    Pennsylvania            0.86487
        21    Texas                   0.82170
        22    Virginia                0.87330
        23    Washington              0.89843
        24    WestVirginia            0.86026
        25    Wisconsin               0.87274


On Wed, May 9, 2012 at 2:53 PM, Price, Joseph <[email protected]> wrote:
> I am running a half-normal production frontier analysis with frontier.
> frontier lnAAA lnBBB  lnCCC lnDDD , distribution(hnormal)  technique(nr)
> predict ehte, te
> predict ehu, u
> predict ehm, m
>
> I am also running a half-normal production frontier analysis with proc QLIM.
> I would like to be able to get the technical efficiencies to match up to STATA using
> frontier. I have been unable to get the same results from both packages.
>
> qlim data=work_dataset1 METHOD=NEWRAP covest=hessian;
> model lnAAA = lnBBB lnCCC lnDDD;
> endogenous lnAAA ~ frontier (type=HALF production);
> output out = work_out1 TE1 TE2 EXPECTED;
>
> The documentation for the technical efficicneies in SAS are
> TE1
> outputs estimates of
> technical efficiency for each producer in the stochastic frontier model
> suggested by Battese and Coelli (1988).
> TE2
> outputs estimates of
> technical efficiency for each producer in the stochastic frontier model
> suggested by Jondrow et al. (1982).
>
> STATA documents
> produces estimates of the technical efficiency via TE = E(exp(-u|e)]
> This has been documented as derrived from (by Battese and Coelli (1988) too?
> and can also produce
> produces estimates of minus the natural log of the technical efficiency via TE = exp(-E(u|e)]
> produces estimates of minus the natural log of the technical efficiency via TE = exp(-M(u|e)]
>
> I would appreciate any help with this issue. Has anyone been able to get the same results from QLIM and frontier?
>
> *
> *   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/

*
*   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