I am performing SFA based on distance functions with STATA. After more than several outputs, I realised that the sign or significance of the parameters may change depending on the order that inputs and outputs are introduced. This also applies for environmental variables in the inefficiency term, the mu may disappear, but if changing the order it comes back again. Is this a known problem, but not acknowledged? Also, how could I increase the number of iterations (likelihood) in STATA from the usual one (100) to maybe 300?I am currently using sfpanel package within STATA. Thank you in advance for your time