I am calculating a stochastic frontier model in STATA with the comand sfpanel. If I use any normal frontier model (such as True Fixed Effects of Greene, 2005) with the same comand I get the coefficient of the signal-to-noise (lambda) parameter, and also its significance. However, when I execute the same model, but including an exogenous factor for the inefficiency term (u), the STATA output does not provide the coefficient nor the significance of lambda. I can calculate the coefficient of lambda as the ratio sigma_u/sigma_v, but how can I calculate the significance?

I have tried using the ereturn list, but it does not provided any information on the lambda parameter. Since I can obtain the inefficiency term, if I were albe to get the error term (v) I could conduct a variance test. However, I have not found any way to estimate the error term with the sfpanel command.

Anybody can help me?

Thanks in anticipation.