Handling losses in translog profit models

Abstract In this paper, we compare standard approaches used to handle losses in logarithmic prot models with a simple novel approach. We estimate translog stochastic prot frontiers, and discuss discriminatory power, rank stability and the precision of prot efciency scores. Contrary to existing methods, our approach does not result in a loss of observations. Our new method enhances rank stability and discriminatory power, and improves the precision of prot efciency scores.