how to predict technical efficiency after estimating a true random effcet model.

[Fissha Asmare]

Dear All,
I am running a true random effcetmodel, which accounts hetroscedasticity using the sfpanel command. But I failed to find the way to predict the technical efficiency(inefficiency). The ususal command predict TE, te is not functional in this case.
Please help me any one who can sort out this.
Thank you!!!

[Scott Merryman]

-sfpanel- is written by F. Belotti, S. Daidone, G. Ilardi, and V. Atella https://www.stata-journal.com/articl...article=st0315

help sfpanel postestimation

Then clicking on -predict- brings up:

    statistic        Description
    ------------------------------------------------------------------------------------------------------
    Main
      xb             linear prediction; the default
      stdp           standard error of the prediction
      u              estimates of (technical or cost) inefficiency via E(u|e) (Jondrow et al., 1982)
      u0             estimates of (technical or cost) inefficiency via E(u|e) when the random effect is
                       zero; only after model(tre)
      m              estimates of (technical or cost) inefficiency via M(u|e)
      jlms           estimates of (technical or cost) efficiency via exp[-E(u|e)]
      bc             estimates of (technical or cost) efficiency via E[exp(-u)|e] (Battese and Coelli,
                       1988)
      ci             estimates of confidence interval for (technical or cost) inefficiency/efficiency
      marginal       marginal effects of the exogenous determinants on the unconditional mean and variance
                       of the inefficiency (Wang, 2002)
      trunc(tlevel)  truncation of estimated efficiency/inefficiency
      scores         calculates score variables

[Fissha Asmare]

Thank you very much Scott. I got it. But I am not still clear with the difference between u, u0, and m and jlms and bc.