Nonparametric frontier estimation from noisy data

摘要: A new nonparametric estimator of production frontiers is defined and studied when the data set units contaminated by measurement error. The error assumed to be an additive normal random variable on input variable, but its variance unknown. a modification m-frontier, which necessitates computation consistent conditional survival function given output variable. In this paper, identification consistency proved in presence noise with unknown variance. performance also through simulated data.