Quantile estimation of the stochastic frontier model

Abstract

The stochastic frontier model remains popular within the field of efficiency analysis and yet it remains deeply connected to the notion of a conditional mean. Recent research has attempted to conceive of, and estimate, the stochastic frontier model in a quantile setting. We demonstrate here that the stochastic frontier corresponds explicitly to a specific quantile of the output distribution and provide a computational approach to recover this quantile. An empirical illustration demonstrates comparable performance with more classical methods of estimation of the stochastic frontier model.