Wrong Skewness and Finite Sample Correction in Parametric Stochastic Frontier Models

In parametric stochastic frontier models, the composed error is specified as the sum of a two-sided noise component and a one-sided inefficiency component, which is usually assumed half-normal, implying that the error distribution is skewed in one direction. In practice, however, estimation residuals may display skewness in the wrong direction. Model re-specification or pulling a new sample is often prescribed. This paper proposes a feasible alternative: imposing a negative skewness constraint on the residuals in maximum likelihood or corrected least squares estimation.