Dear STATA users,

I am estimating the determinant factors of technical efficiency of teff farms (for clarity of my situation; teff is a type of cereal crop, which is the second dominant cereal crop and a staple food in Ethiopia)

I am doing stochastic production frontier analysis on a panel data set (using sfpanel), hence, I am trying to estimate different models like model(tfe), model(tfe), model(bc95) and model(kumb90) with respect to inefficiency distribution (truncated normal, half normal and exponential). Besides, one of them works, the other makes an error (flat or discontinuous region encountered). Also, they gave me different results. As a result, I am so confused to say which model is better for my dataset?
I tried the following codes with translog production function.

sfpanel lnYield lnLand lnOxenQty lnLabor lnSeed lnFertilizer lnLand_lnOxenQty lnLand_lnLabor lnLand_lnSeed lnLand_lnFertilizer lnOxenQty_lnLabor lnOxenQty_lnSeed lnOxenQty_lnFertilizer lnLabor_lnSeed lnLabor_lnFertilizer lnSeed_lnFertilizer _t _t_lnLand _t_lnOxenQty _t_lnLabor _t_lnSeed _t_lnFertilizer _t_2 lnLand_2 lnOxenQty_2 lnLabor_2 lnSeed_2 lnFertilizer_2, model(bc95) ort(0) ..... flat or discontinuous region encountered is occured

sfpanel lnYield lnLand lnOxenQty lnLabor lnSeed lnFertilizer lnLand_lnOxenQty lnLand_lnLabor lnLand_lnSeed lnLand_lnFertilizer lnOxenQty_lnLabor lnOxenQty_lnSeed lnOxenQty_lnFertilizer lnLabor_lnSeed lnLabor_lnFertilizer lnSeed_lnFertilizer _t _t_lnLand _t_lnOxenQty _t_lnLabor _t_lnSeed _t_lnFertilizer _t_2 lnLand_2 lnOxenQty_2 lnLabor_2 lnSeed_2 lnFertilizer_2, model(tfe) dist(hnormal) ort(o) ....... It works

sfpanel lnYield lnLand lnOxenQty lnLabor lnSeed lnFertilizer lnLand_lnOxenQty lnLand_lnLabor lnLand_lnSeed lnLand_lnFertilizer lnOxenQty_lnLabor lnOxenQty_lnSeed lnOxenQty_lnFertilizer lnLabor_lnSeed lnLabor_lnFertilizer lnSeed_lnFertilizer _t _t_lnLand _t_lnOxenQty _t_lnLabor _t_lnSeed _t_lnFertilizer _t_2 lnLand_2 lnOxenQty_2 lnLabor_2 lnSeed_2 lnFertilizer_2 ....... It works but different result.

If anyone has any advice I would very much appreciate it!

Asmiro