CMStatistics 2021: Start Registration
View Submission - CMStatistics
Title: Mixed frontier estimation in the presence of measurement error with unknown variance Authors:  Jun Cai - KU Leuven (Belgium) [presenting]
Ingrid Van Keilegom - KU Leuven (Belgium)
Abstract: Stochastic frontier models for cross-sectional data typically assume that the one-sided distribution of firm-level inefficiency is either continuous or discrete. However, it may be reasonable to hypothesize that inefficiency is continuous except for a discrete mass at zero capturing fully efficient firms (zero-inefficiency). We extend a previous method for such a mixture distribution in the stochastic frontier model with unknown error variance and modify it to incorporate a lower bound frontier estimation as well. Consistency, convergence rates of the estimator are established, as well as a test of the zero-inefficiency hypothesis. Simulations and an application to the cost efficiency of US banks are provided.