Dear all: I am estimating a panel data SFA using the model proposed by Battese and Coelli (1995). I am using the Stata command sfpanel. For the sake of simplicity I explain my problem assuming only one determinant (Z) of both Mu and Usigma. Is the significance of the coefficients of Z a sufficient, necessary or not even necessary condition to derive conclusions about the significance of the marginal effects of Z on the mean and variance of technical efficiency? Maybe it not even a necessary condition. I'd appreciate your help. Thanks, Juan