About postestimation in case of insignificant parameters in stochastic frontier analysis

[Jae-hyung Park]

Dear Statalist,

I performed stochastic frontier analysis by applying Greene's model(2005) with true random effects.
However, I found that some parameters were not significant.

How can I postestimate the (in)efficiency of Decision Making Units?

The code that I used is as follows;

. sfpanel lny lnx charno_dum2 charno_dum3 lnxd2 lnxd3, model(tre)

charno_dum2, charno_dum3 : dummy variables
lnxd2 : interaction term of (independent variable "lnx" and "charno_dum2")
lnxd3 : interaction term of (independent variable "lnx" and "charno_dum3")


And the results are as follows;

Can anyone help me how I can solve this?

Thank you in advance.