Announcement

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

  • Error when estimating Stochastic Frontier model using sfpanel


    Dear Statalist members

    I am trying to estimate a Stochastic Frontier model using the user written command sfpanel, by Belotti et al (2012). Unfortunately I am getting the following error message:

    <istmt>: 3499 _starting_values_cs() not found
    r(3499);

    I got the same error when I tried to run the examples provided in the help file of the command.

    After a search on the internet I found the following link on the Statalist archives offering a solution to the problem.

    http://www.stata.com/statalist/archi.../msg00299.html


    I followed the steps provided by the author of the command, but it did not solve my problem. More specifically I used the following commands and I got the respective errors:

    ado, find(sfcross) - Did not return anything in the Stata results window

    ado uninstall sfpanel and ado uninstall cross - package not found

    ssc unistall sfcross and ssc unistall sfpanel - ssc: unistall: invalid subcommand

    ssc install sfpanel, replace and ssc install sfcross, replace - Returns that all files already exist and are up to date

    Please note that I am using Stata 13.1

    Any help is appreciated


    Belotti et al. (2012). Stochastic frontier analysis using Stata. The Stata Journal, 13(4), 719–758.

    Kind regards
    Aristeidis Dadoukis



  • #2
    It seems that you have outdated versions of the package incorrectly installed. You should remove any ado file (*.ado files) and mata library (*.mlib files) related to -sfcross- and -spanel- from your ~/ado/plus or ~/ado/personal directories and try to reinstall the package using

    net install st0315.pkg
    Federico

    Comment


    • #3
      Hi Federico & Aristeidis!

      Today I installed the sfcross package (package st0315 from http://www.stata-journal.com/software/sj13-4). This should be the up-to-date version, right? However, I face the same problem as Aristeidis. When I run sfcross I receive the error message

      <istmt>: 3499 _starting_values_cs() not found
      r(3499);

      Is there any fix for this? How can my version be outdated, if I directly download it via stata/SJ?

      I further noticed, that the error occurs only when using large data sets. Maybe this information is helpful for you.

      Best regards,
      Benjamin

      BTW I use stata 13.1.

      Comment


      • #4
        Hi all,

        I'm using stata 13.1.

        I observed the same error message using a large Panel data set (<istmt>: 3499 _starting_values_cs() not found
        r(3499)).

        With a smaller data set all went well.

        Is there any possibility to fix this?

        Thanks in advance.

        Best,
        Ben

        Comment


        • #5
          I recently had a program using sfcross working fine but now I get the same error (below) as the others above. This is in STATA 14.

          Any suggestions?
          <istmt>: 3499 _starting_values_cs() not found

          Comment


          • #6
            hi
            I am trying to run following code to estimate cost efficiency while capturing inefficiency through z(exogenous variables) for banks with 143 observations:

            sfpanel lntcost lnx1 lnx2 hflnx1lnx1 lnx1lnx2 hflnx2lnx2 lny1 lny2 lny3 hflny1lny1 lny1lny2 lny1lny3 hflny2lny2 lny2lny3 hflny3lny3 lnx1lny1 lnx1lny2 lnx1lny3 lnx2lny1 lnx2lny2 lnx2lny3 t, model(bc95) usigma(lngdp lnmargn lnadq lnrate lndposit) cost

            STATA run this code but give most of the result's table blank and similarly "predict ce, bce" didn't give any score of cost efficiency.
            Please advice?

            Thanks
            Salah

            Comment


            • #7
              Hello Salah,

              Without an output I cannot be certain, but it seems that your model has not converged. After running the sfpanel, type
              di e(converged)

              If the model has converged, then Stata should return 1.

              If you get 0, then you need to respecify the model or try a different maximization technique (see the maximization techniques offered by Stata)

              Comment

              Working...
              X